[HN Gopher] Relicensing with AI-Assisted Rewrite
___________________________________________________________________
Relicensing with AI-Assisted Rewrite
Author : tuananh
Score : 366 points
Date : 2026-03-05 05:07 UTC (17 hours ago)
(HTM) web link (tuananh.net)
(TXT) w3m dump (tuananh.net)
| verdverm wrote:
| Interesting questions raised by recent SCOTUS refusal to hear
| appeals related to AI an copyright-ability, and how that may
| affect licensing in open source.
|
| Hoping the HN community can bring more color to this, there are
| some members who know about these subjects.
| anilgulecha wrote:
| This is precedent setting. In this case the rewrite was in same
| language, but if there's a python GPL project, and it's tests
| (spec) were used to rewrite specs in rust, and then an
| implementation in rust, can the second project be legally MIT, or
| any other?
|
| If yes, this in a sense allows a path around GPL requirements.
| Linux's MIT version would be out in the next 1-2 years.
| nairboon wrote:
| No, GPL still holds even if you transform the source code from
| one language to another language.
| anilgulecha wrote:
| That why I carved it out to just the specs. If they can be
| read as "facts", then the new code is not derived but arrived
| at with TTD.
|
| The thesis I propose is that tests are more akin to facts, or
| can be stated as facts, and facts are not copyright-able.
| That's what makes this case interesting.
| nairboon wrote:
| I assumed that "tests" refers to a program too, which in
| this example is likely GPL. Thus GPL would stick already on
| the AI-rewrite of GPL test code.
|
| If "tests" should mean a proper specification let's say
| some IETF RFC of a protocol, then that would be different.
| anilgulecha wrote:
| Yes, I had not specified in my original comment. But in
| the SOTA LLM world code/text boundary is so blurry, so as
| to be non-existent.
| mlaretallack wrote:
| Its very important to understand the "how" it was done. The GPL
| hands the "compile" step, and the result is still GPL. The
| clean Room process uses 2 teams, separated by a specification.
| So you would have to
|
| 1. Generate specification on what the system does. 2. Pass to
| another "clean" system 3. Second clean system implements based
| just on the specification, without any information on the
| original.
|
| That 3rd step is the hardest, especially for well known
| projects.
| anilgulecha wrote:
| 1 is claude-code1, outputs tests as text.
|
| 2. Dumped into a file.
|
| 3. claude-code that converts this to tests in the target
| language, and implements the app that passes the tests.
|
| 3 is no longer hard - look at all the reimplementations from
| ccc, to rewrites popping up. They all have a well defined
| test suite as common theme. So much so that tldraw author
| raised a (joke) issue to remove tests from the project.
| microtonal wrote:
| So what if a frontier model company trains two models, one
| including 50% of the world's open source project and the
| second model the other 50% (or ten models with 90-10)?
|
| Then the model that is familiar with the code can write
| specs. The model that does not have knowledge of the project
| can implement them.
|
| Would that be a proper clean room implementation?
|
| Seems like a pretty evil, profitable product "rewrite any
| code base with an inconvenient license to your proprietary
| version, legally".
| anilgulecha wrote:
| LLM training is unnecessary in what we're discussing.
| Merely LLM using: original code -> specs as facts -> specs
| to tests -> tests to new code.
| microtonal wrote:
| It is hard to prove that the model doesn't recognize the
| tests and reproduces the memoized code. It's not a clean
| room.
| yjftsjthsd-h wrote:
| > but if there's a python GPL project, and it's tests (spec)
| were used to rewrite specs in rust, and then an implementation
| in rust, can the second project be legally MIT, or any other?
|
| Isn't that what https://github.com/uutils/coreutils is? GNU
| coreutils spec and test suite, used to produce a rust MIT
| implementation. (Granted, by humans AFAIK)
| hrmtst93837 wrote:
| Treating an AI-assisted rewrite as a legal bypass for the GPL
| is wishful thinking. A defensible path is a documented clean-
| room reimplementation where a team that never saw the GPL
| source writes independent specs and tests, and a separate team
| implements from those specs using black-box characterization
| and differential testing while you document the chain of
| custody.
|
| AI muddies the water because large models trained on public
| repos can reproduce GPL snippets verbatim, so prompting with
| tests that mirror the original risks contamination and a court
| could find substantial similarity. To reduce risk use black-box
| fuzzing and property-based tools, have humans review and scrub
| model outputs, run similarity scans, and budget for legal
| review before calling anything MIT.
| AberrantJ wrote:
| I'm somewhat confused on how it actually muddies the waters -
| any person could have read the source code before hand and
| then either lied about it or forgot.
|
| Our knowledge of what the person or the model actually
| contains regarding the original source is entirely incomplete
| when the entire premise requires there be full knowledge that
| nothing remains.
| Retr0id wrote:
| > In traditional software law, a "clean room" rewrite requires
| two teams
|
| Is the "clean room" process meaningfully backed by legal
| precedent?
| estimator7292 wrote:
| Yes. Compaq's reverse engineering of the IBM PC BIOS set the
| precedent.
| Firehawke wrote:
| Sure. The reimplementation of the IBM PC BIOS that gave birth
| to IBM Compatibles is the canonical example.
| karlding wrote:
| I am not a lawyer, but from my understanding the legal
| precedent is NEC v. Intel which established that clean-room
| software development is not infringing, even if it performs the
| same functionality as the original.
|
| As an aside, this clean room engineering is one of the plot
| points of Season 1 of the TV show Halt and Catch Fire where the
| fictional characters do this with the BIOS image they dumped.
| devmor wrote:
| It is the reason AMD exists.
| DrammBA wrote:
| I like the idea of AI-generated ~code~ _anything_ being public
| domain. Public data in, public domain out.
| lejalv wrote:
| This could be read as a reformulation of the old adage -
| "what's mine is mine, and what is yours, is mine too".
|
| So, you can pilfer the commons ("public") but not stuff
| unavailable in source form.
|
| If we expand your thought experiment to other forms of
| expression, say videos on YT or Netflix, then yes.
| kshri24 wrote:
| I don't think you can classify "public data in" as public
| domain. Public data could also include commercial licenses
| which forbid using it in any way other than what the license
| states. Just because the source is open for viewing does not
| necessarily mean it is OSL.
|
| That's the core issue here. All models are trained on ALL
| source code that is publicly available irrespective of how it
| was licensed. It is illegal but every company training LLMs is
| doing it anyways.
| thedevilslawyer wrote:
| Copyright is not a blacklist but an allowlist of things kept
| aside for the holder. Everything else is free game. LLM
| ingestion comes under fair use so no worries. If someone can
| get their hand on it, nothing in law stops it from training
| ingestion.
|
| We can debate if this law is moral. Like the GP I took agree
| public data in -> public domain out is what's right for
| society. Copyright as an artificial concept has gone on for
| long enough.
| kshri24 wrote:
| > LLM ingestion comes under fair use
|
| I don't think so. It is no where "limited use". Entirety of
| the source code is ingested for training the model. In
| other words, it meets the bar of "heart of the work" being
| used for training. There are other factors as well, such as
| not harming owner's ability to profit from original work.
| thedevilslawyer wrote:
| https://www.skadden.com/insights/publications/2025/07/fai
| r-u...
|
| Both Meta and Anthropic were vindicated for their use.
| Only for Anthropic was their fine for not buying upfront.
| kshri24 wrote:
| This hasn't gone to Supreme Court yet. And this is just
| USA. Courts in rest of the World will also have to take a
| call. It is not as simple as you make it out to be.
| Developers are spread across the World with majority
| living outside USA. Jurisdiction matters in these things.
| thedevilslawyer wrote:
| Copyright's ambit has been pretty much defined and run by
| US for over a century.
|
| You're holding out for some grace on this from the wrong
| venue. The right avenue would be lobbying for new laws to
| regulate and use LLMs, not try to find shelter in an
| archaic and increasingly irrelevant bit of legalese.
| kshri24 wrote:
| I don't disagree. However, just because your assertion of
| copyright being initially defined by US (which is not the
| fact. It was England that came up with it and was adopted
| by the Commonwealth which US was also a part of until its
| independence) does not mean jurisdiction is US. Even if
| US Supreme Court rules one way or the other, it doesn't
| matter as the rest of the World have its own definitions
| and legalese that need to be scrutinized and modernized.
| shakna wrote:
| Alsup absolutely did not vindicate Anthropic as "fair
| use".
|
| > Instead, it was a fair use because all Anthropic did
| was replace the print copies it had purchased for its
| central library with more convenient space-saving and
| searchable digital copies for its central library --
| without adding new copies, creating new works, or
| redistributing existing copies. [0]
|
| It was only fair use, where they already had a license to
| the information at hand.
|
| [0] https://storage.courtlistener.com/recap/gov.uscourts.
| cand.43...
| gf000 wrote:
| There are hardly any rulings/laws about the topic, and it
| quite obviously changes the picture of licenses.
| fschuett wrote:
| > It is illegal
|
| Only (?) in America. In the EU, scraping is legal by default
| unless explicitly opted out with machine-readable
| instructions like robots.txt. That covers "training input".
| For training output, the rule is: "if the output is
| unrecognizable to the input, the license of the input does
| not matter" (otherwise, any project X could sue project Y for
| copyright infringement even if the projects only barely
| resemble each other). The cases where companies actually got
| sued were where the output was a direct copy or repetition of
| the input, even if an LLM was involved.
|
| There is, however, a larger philosophical divide between the
| US and the EU based on history and religion. The US
| philosophy is highly individualistic, capitalistic, and
| considers "first-order principles." Copyright is a "property
| right": "I own this string of bits, you used them, therefore
| you owe me" (principle of absolute ownership).
|
| Continental philosophy is more social and considers "second-
| order / causal effects." Copyright is a "personality right"
| that exists within a social ecosystem. The focus is on the
| effect of the action rather than a singular principle like
| "intellectual property." If the new code provides a secondary
| benefit to society and doesn't "hurt" the original creator's
| unique intellectual stamp, the law is inclined to view it as
| a new work.
|
| In terms of legal sociology, America and Britain are more
| "individual-property-atomistic" thanks to their Protestant
| heritage, focusing on the rights of the individual ( _sola_
| me, and my property, and God). Meanwhile, Europe was, at
| least to a large part, Catholic (esp. France), which focuses
| more on works, results, and effects on society to determine
| morality. While the states are officially secular, the
| heritage of this echoes in different definitions of what is
| considered "legal" or "moral", depending on which side of
| the ocean you are on.
| benob wrote:
| What about doing that with movies and music?
| zodmaner wrote:
| The results would be the same: AI generated music and movies
| will be public domain.
| nkmnz wrote:
| So you'd lose all rights on pictures of yourselves if they
| were generated by AI? Would this be true even for nudes?
| pseudalopex wrote:
| Copyright and privacy rights are different.
| nkmnz wrote:
| I did not refer to privacy rights. If you post a photo of
| yourselves online, you're giving up on a tiny part of
| your privacy rights. So my question still stands: would
| running your photos that you have taken of yourselves
| through a diffusion model rip your copyright of your
| photo?
| DrammBA wrote:
| Yes, _anything_ AI-generated should be public domain
| including the AI-generated picture that used your photo
| as input.
| DannyBee wrote:
| Lawyer here. Its not. This article is highly confused. The case
| was about whether an AI could be considered an author for
| copyright purposes. Mainly as a way of arguing for robot
| rights, not copyright. The person listed the AI as the sole
| author: On the application, Dr. Thaler listed the Creativity
| Machine as the work's sole author and himself as just the
| work's owner.
|
| This is not the first time someone tried to say a machine is
| the author. The law is quite clear, the machine cant be an
| author for copyright purposes. Despite all the confused news
| articles, this does not mean if claude writes code for you it
| is copyright free. It just means you are the author. Machines
| being used as tools to generate works is quite common, even
| autonomously. ill steal from the opinion here:
|
| In 1974, Congress created the National Commission on New
| Technological Uses of Copyrighted Works ("CONTU") to study how
| copyright law should accommodate "the creation of new works by
| the application or intervention of such automatic systems or
| machine reproduction."
|
| ...
|
| This understanding of authorship and computer technology is
| reflected in CONTU's final report: On the basis of its
| investigations and society's experience with the computer, the
| Commission believes that there is no reasonable basis for
| considering that a computer in any way contributes authorship
| to a work produced through its use. The computer, like a camera
| or a typewriter, is an inert instrument, capable of functioning
| only when activated either directly or indirectly by a human.
| When so activated it is capable of doing only what it is
| directed to do in the way it is directed to perform.
|
| ...
|
| IE When you use a computer or any tool you are still the
| author.
|
| The court confirms this later:
|
| Contrary to Dr. Thaler's assumption, adhering to the human-
| authorship requirement does not impede the protection of works
| made with artificial intelligence. Thaler Opening Br. 38-39.
| First, the human authorship requirement does not prohibit
| copyrighting work that was made by or with the assistance of
| artificial intelligence. The rule requires only that the author
| of that work be a human being--the person who created,
| operated, or used artificial intelligence--and not the machine
| itself. The Copyright Office, in fact, has allowed the
| registration of works made by human authors who use artificial
| intelligence.
|
| There are cases where the use of AI made something
| uncopyrightable, even when a human was listed as the author,
| but all of the ones i know are image related.
| DrammBA wrote:
| > Lawyer here. Its not. This article is highly confused.
|
| Did you reply to the wrong comment? I was just saying I like
| the idea of AI-generated _anything_ being public domain, not
| that it currently is /isn't.
| postalrat wrote:
| "the person who created, operated, or used artificial
| intelligence" so which one is it? because there the person(s)
| who created the ai is almost always different that the person
| who used it.
| nairboon wrote:
| That code is still LGPL, it doesn't matter what some release
| engineer writes in the release notes on Github. All original
| authors and copyright holders must have explicitly agreed to
| relicense under a different license, otherwise the code stays
| LGPL licensed.
|
| Also the mentioned SCOTUS decision is concerned with authorship
| of generative AI products. That's very different of this case.
| Here we're talking about a tool that transformed source code and
| somehow magically got rid of copyright due to this
| transformation? Imagine the consequences to the US copyright
| industry if that were actually possible.
| pavlov wrote:
| If anything, the SCOTUS decision would seem to imply that
| generative AI transformations produce no additional creative
| contribution and therefore the original copyright holder has
| all rights to any derived AI works.
|
| (IANAL)
| dathinab wrote:
| that is a very good formulation of what I have been trying to
| say
|
| but also probably not fully right
|
| as far as I understand they avoid the decision of weather an
| AI can produce creative work by saying that the neither the
| AI nor it's owner/operator can claim ownership of copyright
| (which makes it de-facto public domain)
|
| this wouldn't change anything wrt. derived work still having
| the original authors copyright
|
| but it could change things wrt. parts in the derived work
| which by themself are not derived
| pseudalopex wrote:
| The court avoided a decision of what the operator could
| have copyrighted because he said he was not the author.
| bandrami wrote:
| That's a reasonable theory though it's stuck with the problem
| that any model will by its training be derivative of
| codebases that have incompatible licenses, and that in fact
| every single use of an LLM is therefore illegal (or at least
| tortious).
| dathinab wrote:
| iff it went through the full clean room rewrite just using AI
| then no, it's de-facto public domain (but also it probably
| didn't do so)
|
| iff it is a complete new implementation with completely
| different internal then it could also still be no LGPL even if
| produced by a person with in depth knowledge. Copyright only
| cares if you "copied" something not if you had "knowledge" or
| if it "behaves the same". So as long as it's distinct enough it
| can still be legally fine. The "full clean room" requirement is
| about "what is guaranteed to hold up in front of a court" not
| "what might pass as non-derivative but with legal risk".
| pocksuppet wrote:
| In the legal system there's no such thing as "code that is
| LGPL". It's not an xattr attached to the code.
|
| There is an act of copying, and there is whether or not that
| copying was permitted under copyright law. If the author of the
| code said you can copy, then you can. If the original author
| didn't, but the author of a derivative work, who wasn't allowed
| to create a derivative work, told you you could copy it, then
| it's complicated.
|
| And none of it's enforced except in lawsuits. If your work was
| copied without permission, you have to sue the person who did
| that, or else nothing happens to them.
| foota wrote:
| I think the more interesting question here would be if someone
| could fine tune an open weight model to remove knowledge of a
| particular library (not sure how you'd do that, but maybe
| possible?) and then try to get it to produce a clean room
| implementation.
| benob wrote:
| I don't think this would qualify as clean room (the Library was
| involved in learning to generate programs as a whole). However,
| it should be possible to remove the library from the OLMO
| training data and retrain it from scratch.
|
| But what about training without having seen any human written
| program? Coul a model learn from randomly generated programs?
| foota wrote:
| > I don't think this would qualify as clean room (the Library
| was involved in learning to generate programs as a whole)
|
| Hm... I mean this is really one for the lawyers, but IMO you
| would likely successfully be able to argue that the marginal
| knowledge of general coding from a particular library is
| likely close to nil.
|
| The hard part here imo would be convincingly arguing that you
| can wipe out knowledge of the library from the training set,
| whether through fine tuning or trying to exclude it from the
| dataset.
|
| > But what about training without having seen any human
| written program? Coul a model learn from randomly generated
| programs?
|
| I think the answer at this point is definitely no, but maybe
| someday. I think it's a more interesting question for art
| since it's more subjective, if we eventually get to a point
| where a machine can self-teach itself art from nothing...
| first of all how, but second of all it would be interesting
| to see the reaction from people opposed to AI art on the
| basis of it training off of artists.
|
| Honestly given all I've seen models do, I wouldn't be too
| surprised if you could somehow distill a (very bad) image
| generation model off of just an LLM. In a sense this is the
| end goal of the pelican riding a bicycle (somewhat tongue in
| cheek), if the LLM can learn to draw anything with SVGs
| without ever getting visual inputs then it would be very
| interesting :)
| kshri24 wrote:
| > The ownership void: If the code is truly a "new" work created
| by a machine, it might technically be in the public domain the
| moment it's generated, rendering the MIT license moot.
|
| How would that work? We still have no legal conclusion on whether
| AI model generated code, that is trained on all publicly
| available source (irrespective of type of license), is legal or
| not. IANAL but IMHO it is totally illegal as no permission was
| sought from authors of source code the models were trained on. So
| there is no way to just release the code created by a machine
| into public domain without knowing how the model was inspired to
| come up with the generated code in the first place. Pretty sure
| it would be considered in the scope of "reverse engineering" and
| that is not specific only to humans. You can extend it to
| machines as well.
|
| EDIT: I would go so far as to say the most restrictive license
| that the model is trained on should be applied to all model
| generated code. And a licensing model with original authors (all
| Github users who contributed code in some form) should be setup
| to be reimbursed by AI companies. In other words, a % of profits
| must flow back to community as a whole every time code-related
| tokens are generated. Even if everyone receives pennies it
| doesn't matter. That is fair. Also should extend to artists whose
| art was used for training.
| thedevilslawyer wrote:
| That's unpractical enough that you might as well wish for UBI
| and world peace rather than this.
| kshri24 wrote:
| Why is it impractical? Github already has a sponsor system.
| Also this can be a form of UBI.
| adrianN wrote:
| We'll have to wait until the technology progresses sufficiently
| that AI cuts into Disney's profit.
| kouteiheika wrote:
| > I would go so far as to say the most restrictive license that
| the model is trained on should be applied to all model
| generated code.
|
| That license is called "All Rights Reserved", in which case you
| wouldn't be able to legally use the output for anything.
|
| There are research models out there which are trained on only
| permissively licensed data (i.e. no "All Rights Reserved"
| data), but they're, colloquially speaking, dumb as bricks when
| compared to state-of-art.
|
| But I guess the funniest consequence of the "model outputs are
| a derivative work of their training data" would be that it'd
| essentially wipe out (or at very least force a revert to a pre-
| AI era commit) every open source project which may have
| included any AI-generated or AI-assisted code, which currently
| pretty much includes every major open source project out there.
| And it would also make it impossible to legally train any new
| models whose training data isn't strictly pre-AI, since
| otherwise you wouldn't know whether your training data is
| contaminated or not.
| kshri24 wrote:
| I agree with your assessment. Which is why I was proposing a
| middle-ground where an agreement is setup between the model
| training company and the collective of developers/artists et
| all and come up with a license agreement where they are
| rewarded for their original work for perpetuity. A tiny % of
| the profits can be shared, which would be a form of UBI. This
| is fair not only because companies are using AI generated
| output but developers themselves are also paying and using AI
| generated output that is trained on other developer's input.
| I would feel good (in my conscience) that I am not "stealing"
| someone else's effort and they are being paid for it.
| kouteiheika wrote:
| > Which is why I was proposing a middle-ground where an
| agreement is setup between the model training company and
| the collective of developers/artists et all and come up
| with a license agreement where they are rewarded for their
| original work for perpetuity. A tiny % of the profits can
| be shared, which would be a form of UBI. This is fair
|
| That wouldn't be fair because these models are not only
| trained on code. A huge chunk of the training data are just
| "random" webpages scraped off the Internet. How do you
| propose those people are compensated in such a scheme? How
| do you even know who contributed, and how much, and to whom
| to even direct the money?
|
| I think the only "fair" model would be to essentially
| require models trained on data that you didn't explicitly
| license to be released as open weights under a permissive
| license (possibly with a slight delay to allow you to
| recoup costs). That is: if you want to gobble up the whole
| Internet to train your model without asking for permission
| then you're free to do so, but you need to release the
| resulting model so that the whole humanity can benefit from
| it, instead of monopolizing it behind an API paywall like
| e.g. OpenAI or Anthropic does.
|
| Those big LLM companies harvest everyone's data en-masse
| without permission, train their models on it, and then not
| only they don't release jack squat, but have the gall to
| put up malicious explicit roadblocks (hiding CoT traces,
| banning competitors, etc.) so that no one else can do it
| _to them_ , and when people try they call it an
| "attack"[1]. _This_ is what people should be angry about.
|
| [1] -- https://www.anthropic.com/news/detecting-and-
| preventing-dist...
| duskdozer wrote:
| >under a permissive license
|
| well, assuming all data that is itself not permissively
| licensed is excluded
| carlob wrote:
| Why settle on some private agreement between creators and
| ai companies where a tiny percentage is shared, let's just
| tax the hell out of AI companies and redistribute.
| kshri24 wrote:
| > let's just tax the hell out of AI companies and
| redistribute.
|
| That's not what I favor because you are inserting a
| middleman, the Government, into the mix. The Government
| ALWAYS wants to maximize tax collections AND fully
| utilize its budget. There is no concept of "savings" in
| any Government anywhere in the World. And Government
| spending is ALWAYS wasteful. Tenders floated by
| Government will ALWAYS go to companies that have
| senators/ministers/prime ministers/presidents/kings etc
| as shareholders. In other words, the tax money collected
| will be redistributed again amongst the top 500
| companies. There is no trickle down. Which is why
| agreements need to be between creators and those who are
| enjoying fruits of the creation. What have Governments
| ever created except for laws that stifle
| innovation/progress every single time?
| LadyCailin wrote:
| Uh, no? https://en.wikipedia.org/wiki/Government_Pension_
| Fund_of_Nor...
|
| Just because you have a failure of imagination for how
| government should work, doesn't mean it can't work. And
| stifling innovation is exactly what I want, when that
| innovation is "steal from everyone so we can invent the
| torment nexus" or whatever's going on these days.
| kshri24 wrote:
| Pension fund is an example of what exactly? All countries
| have pension funds. This has nothing to do with
| Governments wasting money. Please go beyond tiny European
| countries that have very few verticals and are largely
| dependent on outside support for protecting their
| sovereignty. They are not representative of most of the
| World.
|
| > As its name suggests, the Government Pension Fund
| Global is invested in international financial markets, so
| the risk is independent from the Norwegian economy. The
| fund is invested in 8,763 companies in 71 countries (as
| of 2024).
|
| Basically what I said above. You give your tax dollars to
| Government and it will invest it into top 500 companies.
| In the Norway Pension Fund case it is 8,763 companies in
| 71 countries. None of them are startups/small
| businesses/creators.
|
| > And stifling innovation is exactly what I want, when
| that innovation is "steal from everyone so we can invent
| the torment nexus" or whatever's going on these days.
|
| You are confusing current lack of laws regulating this
| space with innovation being evil. Innovation is not evil.
| The technology per se is not evil. Every innovation
| brings with it a set of challenges which requires us to
| think of new legislation. This has ALWAYS been the case
| for thousands of years of human innovation.
| carlob wrote:
| > What have Governments ever created except for laws that
| stifle innovation/progress every single time?
|
| https://www.youtube.com/watch?v=Qc7HmhrgTuQ
|
| In all seriousness without the government you would have
| no innovation and progress, because it's the public
| school system, functioning roads, research grants a
| stable and lawful society that allow you to do any kind
| of innovation.
|
| Apart from that, you have answered to a strawman. I said
| redistribute, not give to the government. I explicitly
| worded things that way because I don't think we should
| not be having a discussion on policy.
|
| I think we are moving to an economy where the share of
| profits taken by capital becomes much larger than the one
| take from labor. If that happens then laborers will have
| very little discretionary income to fuel consumption and
| even capitalists will end up suffering. We can choose to
| redistribute now or wait for it to happen naturally,
| however that usually happens in a much more violent way,
| be it hyperinflation, famine, war or revolution.
| kshri24 wrote:
| > Apart from that, you have answered to a strawman. I
| said redistribute, not give to the government
|
| You said: "let's just tax the hell out of AI companies
| and redistribute.". Only the Government has the power to
| tax. Question of redistribution does not even arise
| without first having the power to the coffers of the
| Company. Which you nor I have. Government CAN have if it
| wants to by either Nationalizing the Company or as you
| said "taxing the hell out of" the company. Please explain
| how you would go about taxing and redistributing without
| involving the Government?
|
| > In all seriousness without the government you would
| have no innovation and progress, because it's the public
| school system, functioning roads, research grants a
| stable and lawful society that allow you to do any kind
| of innovation.
|
| These fall under the ambit of governance and hence why
| you have a Government. That's the only power Governments
| should have. Governments SHOULD NOT be managing private
| enterprises.
|
| > I think we are moving to an economy where the share of
| profits taken by capital becomes much larger than the one
| take from labor. If that happens then laborers will have
| very little discretionary income to fuel consumption and
| even capitalists will end up suffering. We can choose to
| redistribute now or wait for it to happen naturally,
| however that usually happens in a much more violent way,
| be it hyperinflation, famine, war or revolution.
|
| Agreed. Which is why I was proposing private agreements
| in the first place (without involving a third-party like
| the Government which, more often than not, mismanages
| funds).
| rswail wrote:
| Because the authors of the original content deserve
| recompense for their work.
|
| That's what the whole copyright and patent regimes are
| designed to achieve.
|
| It's to encourage the creation of knowledge.
|
| US Constitution, Article I, section 8:
| To promote the Progress of Science and useful Arts, by
| securing for limited Times to Authors and Inventors the
| exclusive Right to their respective Writings and
| Discoveries;
| carlob wrote:
| Right, it says exclusive rights, which does not translate
| to "we siphon everything and you get a tiny percentage of
| our profits", it means I can choose to say no to all of
| this. To me the matter of compensation and that of
| authorship rights are mostly orthogonal.
| progval wrote:
| > There are research models out there which are trained on
| only permissively licensed data
|
| Models whose authors _tried_ to train only on permissively
| licensed data.
|
| For example https://huggingface.co/bigcode/starcoder2-15b
| tried to be a permissively licensed dataset, but it filtered
| only on repository-level license, not file-level. So when
| searching for "under the terms of the GNU General Public
| License" on https://huggingface.co/spaces/bigcode/search-v2
| back when it was working, you would find it was trained on
| many files with a GPL header.
| foota wrote:
| I don't know how far it would get, but I imagine that a FAANG
| will be able to get the farthest here by virtue of having
| mountains of corporate data that they have complete ownership
| over.
| msdz wrote:
| They'd probably get the farthest, but they won't pursue
| that because they don't want to end up leaking the original
| data from training. It is possible in regular language/text
| subsets of models to reconstruct massive consecutive parts
| of the training data [1], so it ought to be possible for
| their internal code, too.
|
| [1] https://arxiv.org/abs/2601.02671
| foota wrote:
| Copyright for me not for thee? :) That's a good point
| though. Maybe they could round trip things? E.g., use the
| model trained only on internal content to generate
| training data (which you could probably do some kind of
| screening to remove anything you don't want leaking) and
| then train a new model off just that?
| pocksuppet wrote:
| More precisely, "All Rights Reserved" is the explicit lack of
| any license.
| d1sxeyes wrote:
| > We still have no legal conclusion on whether AI model
| generated code, that is trained on all publicly available
| source (irrespective of type of license), is legal or not.
|
| That horse has bolted. No one knows where all the AI code any
| more, and it would no longer possible to be compliant with a
| ruling that no one can use AI generated code.
|
| There may be some mental and legal gymnastics to make it
| possible, but it will be made legal because it's too late to do
| anything else now.
| conartist6 wrote:
| I hate that this may be true, but I also don't think the law
| will fix this for us.
|
| I think this is down the community and the culture to draw
| our red lines on and enforce them. If we value open source,
| we will find a way to prevent its complete collapse through
| model-assisted copyright laundering. If not, OSS will be
| slowly enshittified as control of projects slowly flows to
| the most profit-motivated entities.
| d1sxeyes wrote:
| But what tools do we have to stop this happening? I agree,
| we can (and should) all refuse to participate in licence
| laundering, but there will always be folks less principled.
|
| I don't know what happens next, honestly.
| conartist6 wrote:
| I don't either, but I guess we're both about to find out.
| There only surety is that there will be moves and
| countermoves. As far as I could tell the best thing we
| could do right now is fund software-legal organizations
| like the EFF which are likely to be the ones to litigate
| the test cases. What's hurting us most right now is we
| don't know what law means in this context, so we don't
| fully understand the scale of what we need to protect
| against or what tools we have that the courts will
| recognize
| shevy-java wrote:
| "We still have no legal conclusion on whether AI model
| generated code, that is trained on all publicly available
| source (irrespective of type of license), is legal or not."
|
| I think it will depend on the way HOW the AI arrived to the new
| code.
|
| If it was using the original source code then it probably is
| guilty-by-association. But in theory an AI model could also
| generate a rewrite if being fed intermediary data not based on
| that project.
| amelius wrote:
| You should just look at it as a giant computation graph. If
| some of the inputs in this graph are tainted by copyright and
| an output depends on these inputs (changing them can change
| the output) then the output is tainted too.
| dathinab wrote:
| > "We still have no legal conclusion on whether AI model
| generated code, that is trained on all publicly available
| source (irrespective of type of license), is legal or not."
|
| it depends on the country you are in
|
| but overall in the US judges have mostly consistently ruled
| it as legal
|
| and this is extremely unlikely to change/be effectively
| interpreted different
|
| but where things are more complex is:
|
| - model containing training data (instead of generic
| abstractions based on it), determined by weather or not it
| can be convinced to produce close to verbatim output of the
| training data the discussion is about
|
| - model producing close to verbatim training data
|
| the later seems to be mostly? always? be seen as copyright
| violation, with the issue that the person who does the
| violation (i.e. uses the produced output) might not known
|
| the former could mean that not just the output but the model
| itself can count as a form of database containing copyright
| violating content. In which case they model provider has to
| remove it, which is technically impossible(1)... The pain
| point with that approach is that it will likely kill public
| models, while privately kept models will for every case put
| in a filter and _claim_ to have removed it and likely will
| get away with it. So while IMHO it should be a violation
| conceptually, it probably is better if it isn't.
|
| But also the case the original article refers to is more
| about models interacting/using with code base then them being
| trained on.
|
| (1): For LLMs, it is very much removable for knowledge based
| used by LLMs.
| dathinab wrote:
| > how does that work
|
| AI can't claim ownership, humans can't either as they haven't
| produced it. If there is guaranteed no one which can claim
| ownership it often seen as being in the public domain.
|
| In general it is irrelevant what the copyright of the AI
| training data is. At least in the US judges have been relevant
| clear about that. (Except if the AI reproduced input data close
| to verbatim. _But in general we aren't speaking about AI being
| trained on a code base but an AI using/rewriting it_.)
|
| (1): Which isn't the same as no one seems to know who has
| ownership. It also might be owned by no-one in the sense that
| no one can grant you can copyright permission (so opposite of
| public domain), but also no-one can sue (so de-facto public
| domain).
| jacquesm wrote:
| Humans can't claim ownership, but they are still liable for
| the product of their bot. That's why MS was so quick to
| indemnify their users, they know full well that it is going
| to be super hard to prove that there is a key link to some
| original work.
|
| The main analogy is this one: you take a massive pile of
| copyrighted works, cut them up into small sections and toss
| the whole thing in a centrifuge, then, when prompted to
| produce a work you use a statistical method to pull pieces of
| those copyrighted works out of the centrifuge. Sometimes you
| may find that you are pulling pieces out of the laundromat in
| the order in which they went in, which after a certain number
| of tokens becomes a copyright violation.
|
| This suggests there are some obvious ways in which AI
| companies can protect themselves from claims of infringement
| but as far as I'm aware not a single one has protections in
| place to ensure that they do not materially reproduce any
| fraction of the input texts other than that they recognize
| prompts asking it to do so.
|
| So it won't produce the lyrics of 'Let it be'. But they'll be
| happy to write you mountains of prose that strongly resembles
| some of the inputs.
|
| The fact that they are not doing that tells you all you
| really need to know: they know that everything that their
| bots spit out is technically derived from copyrighted works.
| They also have armies of lawyers and technical arguments to
| claim the opposite.
| dathinab wrote:
| > Humans can't claim ownership, but they are still liable
| for the product of their bot.
|
| sure,
|
| but that is completely unrelated to this discussion
|
| which is about AI using code as input to produce similar
| code as output
|
| not about AI being trained on code
| jacquesm wrote:
| > which is about AI using code as input to produce
| similar code as output
|
| > not about AI being trained on code
|
| The two are very directly connected.
|
| The LLM would not be able to do what it does without
| being trained, and it was trained on copyrighted works of
| others. Giving it a piece of code for a rewrite is a
| clear case of transformation, no matter what, but now it
| _also_ rests on a mountain of other copyrighted code.
|
| So now you're doubly in the wrong, you are willfully
| using AI to violate copyright. AI does not create
| original works, period.
| bluGill wrote:
| Every programmer is trained on the copyrighted works of
| others. there a vanishingly few modern programs with
| available source code in the public domain.
|
| it isn't clear how/if llm is different from the brain but
| we all have training by looking at copywrited source code
| at some time.
| jacquesm wrote:
| > it isn't clear how/if llm is different from the brain
|
| It's very clear: the one is a box full of electronics,
| the other is part of the central nervous system of a
| human being.
|
| > but we all have training by looking at copywrited
| source code at some time.
|
| That may be so, but not usually the copyrighted source
| code that we are trying to reproduce. And that's the bit
| that matters.
|
| You can attempt to whitewash it but at its core it is
| copyright infringement and the creation of derived works.
| SiempreViernes wrote:
| > but we all have training by looking at copywrited[sic]
| source code at some time.
|
| The single word "training" is here being used to describe
| two very different processes; what an LLM does with text
| during training is at basically every step fundamentally
| distinct from what a human does with text.
|
| Word embedding and gradient descent just aren't anything
| at all like reading text!
| jacquesm wrote:
| Indeed, but that's just a misdirection. We don't actually
| know how a human brain learns, so it is hard to base any
| kind of legal definition on that difference. Obviously
| there are massive differences but what those differences
| are is something you can debate just about forever.
|
| I have a lot of music in my head that I've listened to
| for decades. I could probably replicate it note-for-note
| given the right gear and enough time. But that would not
| make any of my output copyrightable works. But if I
| doodle for three minutes on the piano, even if it is
| going to be terrible that _is_ an original work.
| pocksuppet wrote:
| Programmer training and AI training are legally distinct
| processes.
| graemep wrote:
| > humans can't either as they haven't produced it. If there
| is guaranteed no one which can claim ownership it often seen
| as being in the public domain.
|
| Says who?. The US ruling the article refers to does not cover
| this.
|
| It is different in other countries. Even if US law says it is
| public domain (which is probably not the case) you had better
| not distribute it internationally. For example, UK law
| explicitly says a human is the author of machine generated
| content: https://news.ycombinator.com/item?id=47260110
| m4rtink wrote:
| I would be totally fine with all code generated by LLMs being
| considered to be under GPL v3 unless the model authors can
| prove without any doubt it was not trained on any GPL v3 code -
| viral licensing to the max. ;-)
| mfabbri77 wrote:
| This has the potential to kill open source, or at least the most
| restrictive licenses (GPL, AGPL, ...): if a license no longer
| protects software from unwanted use, the only possible strategy
| is to make the development closed source.
| _dwt wrote:
| Yes, this is the reason I've completely stopped releasing any
| open-source projects. I'm discovering that newer models are
| somewhat capable of reverse-engineering even compiled
| WebAssembly, etc. too, so I can feel a sort of "dark forest
| theory" taking hold. Why publish anything - open or closed - to
| be ripped off at negligible marginal cost?
| seddonm1 wrote:
| It's a real problem. I threw it at an old MUD game just to
| see how hard it is [0] then used differential testing and
| LLMs to rewrite it [1]. Just seems to be time and money.
|
| [0] https://reorchestrate.com/posts/your-binary-is-no-longer-
| saf...
|
| [1] https://reorchestrate.com/posts/your-binary-is-no-longer-
| saf...
| evanelias wrote:
| Wow, as a former MajorMUD addict (~30 years ago) that's
| extremely interesting to see. Especially since MajorMUD is
| rarely discussed on HN, even in MUD or BBS-related threads.
|
| Did you find it worked reasonably well on any portion of
| the codebase you could throw at it? For example, if I
| recall correctly, all of MajorMUD's data file interactions
| used the embedded Btrieve library which was popular at the
| time. For that type of specialized low-level library, I'm
| curious how much effort it would take to get readable code.
| Tiberium wrote:
| People are just not realizing this now because it's mostly
| hobby projects and companies doing it in private, but
| eventually everyone will realize that LLMs allow almost any
| software to be reverse engineered for cheap.
|
| See e.g. https://banteg.xyz/posts/crimsonland/ , a single
| human with the help of LLMs reverse engineered a non-trivial
| game and rewrote it in another language + graphics lib in 2
| weeks.
| abrookewood wrote:
| Why does it matter if it is 'ripped off' if you released it
| as open source anyway? I get that you might want to impose a
| particular licence, but is that the only reason?
| vbarrielle wrote:
| Even the most permissive open source licenses such as MIT
| require attribution. Releasing as open source would
| therefore benefit the author through publicity. Bein able
| to say that you're the author of library X, used by
| megacorp Y with great success, is a good selling point in a
| job interview.
|
| LLM ripping off open source code removes that.
| abrookewood wrote:
| Yes, that's a good point.
| bogwog wrote:
| This is pretty much exactly why copyright laws came about in
| the first place. Why bother creating a book, painting, or
| other work of art if anyone can trivially copy it and sell it
| without handing you a dime?
|
| I think refusing to publish open source code right now is the
| safe bet. I know I won't be publishing anything new until
| this gets definitively resolved, and will only limit myself
| to contributing to a handful of existing open source
| projects.
| abrookewood wrote:
| It's not just open source, it is literally anything source-
| available, whether intentional or not.
| user34283 wrote:
| I find the wording "protect from unwanted use" interesting.
|
| It is my understanding that what a GPL license requires is
| releasing the source code of modifications.
|
| So if we assume that a rewrite using AI retains the GPL
| license, it only means the rewrite needs to be open source
| under the GPL too.
|
| It doesn't prevent any unwanted use, or at least that is my
| understanding. I guess unwanted use in this case could mean not
| releasing the modifications.
| mfabbri77 wrote:
| If the AI product is recognised as "derivative work" of a
| GPL-compliant project, then it must itself be licensed under
| the GPL. Otherwise, it can be licensed under any other
| license (including closed source/proprietary binary
| licenses). This last option is what threatens to kill open
| source: an author no longer has control over their project.
| This might work for permissive licenses, but for GPL/AGPL and
| similar licenses, it's precisely the main reason they exist:
| to prevent the code from being taken, modified, and treated
| as closed source (including possible use as part of
| commercial products or Sass).
| duskdozer wrote:
| Yeah, the GPL is deficient in that way and doesn't handle
| other hostile uses.
| GaryBluto wrote:
| If you'd be willing to close source your "libre" open source
| project because somebody might do something you don't like with
| it, you never wanted a "libre" project.
| saagarjha wrote:
| In this case someone is making a non-libre project with it.
| himata4113 wrote:
| I mean in my opinion GPL licensed code should just infect models
| forcing them to follow the license.
|
| You can do this a lot by saying things like: complete the code
| "<snippet from gpl licensed code>".
|
| And if now the models are GPL licensed the problem of relicensing
| is gone since the code produced by these models should in theory
| be also GPL licensed.
|
| Unfortunately, there is a dumb clause that computer generated
| code cannot be copyrighted or licensed to begin with.
| kshri24 wrote:
| > Unfortunately, there is a dumb clause that computer generated
| code cannot be copyrighted or licensed to begin with.
|
| Can you point to the clause? I have never seen it in any GPL
| license.
| himata4113 wrote:
| it's the general copyright protection 'law' fair use and all
| that, varies by country tho.
| Tomte wrote:
| > The original author, a2mark , saw this as a potential GPL
| violation
|
| Mark Pilgrim! Now that's a name I haven't read in a long time.
| samrus wrote:
| > The ownership void: If the code is truly a "new" work created
| by a machine, it might technically be in the public domain the
| moment it's generated, rendering the MIT license moot.
|
| Im struggling to see where this conclusion came from. To me it
| sounds like the AI-written work can not be coppywritten, and so
| its kind of like a copy pasting the original code. Copy pasting
| the original code doesnt make it public domain. Ai gen code cant
| be copywritten, or entered into the public domain, or used for
| purposes outside of the original code's license. Whats the
| paradox here?
| laksjhdlka wrote:
| They say "if" it's a new work, then it might not be
| copyrightable, I guess. You suppose that it's still the
| original work, and hence it's still got that copyright.
|
| I think they are rhetorically asking if your position is
| correct.
| NitpickLawyer wrote:
| > To me it sounds like the AI-written work can not be
| coppywritten
|
| I think we didn't even began to consider all the implications
| of this, and while people ran with that one case where someone
| couldn't copyright a generated image, it's not that easy for
| code. I think there needs to be way more litigation before we
| can confidently say it's settled.
|
| If "generated" code is not copyrightable, where do draw the
| line on what generated means? Do macros count? Does code that
| generates other code count? Protobuf?
|
| If it's the tool that generates the code, again where do we
| draw the line? Is it just using 3rd party tools? Would training
| your own count? Would a "random" code gen and pick the winners
| (by whatever means) count? Bruteforce all the space (silly
| example but hey we're in silly space here) counts?
|
| Is it just "AI" adjacent that isn't copyrightable? If so how do
| you define AI? Does autocomplete count? Intellisense? Smarter
| intellisense?
|
| Are we gonna have to have a trial where there's at least one
| lawyer making silly comparisons between LLMs and power plugs?
| Or maybe counting abacuses (abaci?)... "But your honour, it's
| just random numbers / matrix multiplications...
| lelanthran wrote:
| All of your questions have seemingly trivial answers. Maybe I
| am missing something, but...
|
| > If "generated" code is not copyrightable, where do draw the
| line on what generated means? Do macros count?
|
| Does the output of the macro depend on ingesting someone
| else's code?
|
| > Does code that generates other code count?
|
| Does the output of the code depend on ingesting someone
| else's code?
|
| > Protobuf?
|
| Does your protobuf implementation depend on ingesting someone
| else's code?
|
| > If it's the tool that generates the code, again where do we
| draw the line?
|
| Does the tool depend ingestion of of someone else's code?
|
| > Is it just using 3rd party tools?
|
| Does the 3rd party tool depend on ingestion of someone else's
| code?
|
| > Would training your own count?
|
| Does the training ingest someone else's code?
|
| > Would a "random" code gen and pick the winners (by whatever
| means) count?
|
| Does the random codegen depend on ingesting someone else's
| code?
|
| > Bruteforce all the space (silly example but hey we're in
| silly space here) counts?
|
| Does the bruteforce algo depend on ingesting someone else's
| code?
|
| > Is it just "AI" adjacent that isn't copyrightable?
|
| No, it's the "depends on ingesting someone else's code" that
| makes it not copyrightable.
|
| > If so how do you define AI?
|
| Doesn't matter whether it is AI or not, the question is are
| you ingesting someone else's code.
|
| > Does autocomplete count?
|
| Does the specific autocomplete in question depend on
| ingesting someone else's code?
|
| > Intellisense?
|
| Does the specific Intellisense in question depend on
| ingesting someone else's code?
|
| > Smarter intellisense?
|
| Does the specific Smarter Intellisense in question depend on
| ingesting someone else's code?
|
| ...
|
| Look, I see where you're going with this - reductio ad
| absurdum and all - but it seems to me that you're trying to
| muddy the waters by claiming that either _all code generation
| is allowed_ or _no code generation is disallowed_.
|
| Let me clear the waters for all the readers - the complaint
| is not about code generation, it's about _ingesting someone
| else 's code_, frequently for profit.
|
| All these questions you are asking seem to me to be
| irrelevant and designed to shift the focus from the ingestion
| of other people's work to something that no one is arguing
| against.
| NitpickLawyer wrote:
| Interesting.
|
| > the complaint is not about code generation, it's about
| ingesting someone else's code, frequently for profit.
|
| Why do you think that is, and _what_ complaint
| specifically? I was talking about this:
|
| > The Copyright Office reviewed the decision in 2022 and
| determined that the image doesn't include "human
| authorship," disqualifying it from copyright protection
|
| There seems to be 0 mentioning of training there. In fact
| if you read the appeal's court case [1] they don't mention
| training either:
|
| > We affirm the denial of Dr. Thaler's copyright
| application. The Creativity Machine cannot be the
| recognized author of a copyrighted work because the
| Copyright Act of 1976 requires all eligible work to be
| authored in the first instance by a human being. Given that
| holding, we need not address the Copyright Office's
| argument that the Constitution itself requires human
| authorship of all copyrighted material. Nor do we reach Dr.
| Thaler's argument that he is the work's author by virtue of
| making and using the Creativity Machine because that
| argument was waived before the agency.
|
| I have no idea where you got the idea that this was about
| training data. Neither the copyright office nor the appeals
| court even mention this.
|
| But anyway, since we're here, let's entertain this. So
| you're saying that training data is the differentiator. OK.
| So in that case, would training on "your own data" make
| this ok with you? Would training on "synthetic" data be ok?
| Would a model that sees no "proprietary" code be ok? Would
| a hypothetical model trained just on RL with nothing but a
| compiler and endless compute be ok?
|
| The courts seem to hint that "human authorship" is still
| required. I see no end to the "... but what about x", as I
| stated in my first comment. I was honestly asking those
| questions, because the crux of the case here rests on
| "human authorship of the piece to be copyrighted", not on
| anything prior.
|
| [1] - https://fingfx.thomsonreuters.com/gfx/legaldocs/egpbl
| okwqpq/...
| lelanthran wrote:
| > There seems to be 0 mentioning of training there. In
| fact if you read the appeal's court case [1] they don't
| mention training either:
|
| > ...
|
| > I have no idea where you got the idea that this was
| about training data. Neither the copyright office nor the
| appeals court even mention this.
|
| In both the story _and_ the comments, that 's the
| prevailing complaint. FTFA:
|
| > Their claim that it is a "complete rewrite" is
| irrelevant, since they had ample exposure to the
| originally licensed code (i.e. this is not a "clean room"
| implementation). Adding a fancy code generator into the
| mix does not somehow grant them any additional rights.
|
| I mean, I know it's passe to read the story, but I still
| do it so my comments are on the story, not just the title
| taken out of context.
|
| > But anyway, since we're here, let's entertain this. So
| you're saying that training data is the differentiator.
|
| Well, that's the complaint in the story and in the
| comment section, so it makes sense to address that and
| that alone.
|
| > OK. So in that case, would training on "your own data"
| make this ok with you?
|
| Yes.
|
| > Would training on "synthetic" data be ok?
|
| If provenance of "synthetic data" does not depend on some
| upstream ingesting someone else's work, then yes.
|
| > Would a model that sees no "proprietary" code be ok?
|
| If the model does not depend on someone else's work, then
| Yes.
|
| > Would a hypothetical model trained just on RL with
| nothing but a compiler and endless compute be ok?
|
| Yes.
|
| *Note: Let me clarify that "someone else's work" means
| someone who has not consented or licended their work for
| ingestion and subsequent reproduction under the terms
| that AI/LLM training does it. If someone licensed you
| their work to train a model, then have at it.
| NitpickLawyer wrote:
| Ah! I think I get where the confusion was. I was quoting
| something from another comment, and specifically
| commenting on that.
|
| > > To me it sounds like the AI-written work can not be
| coppywritten
|
| I was only commenting on that.
| user34283 wrote:
| I'm thinking that the relevant question would be whether
| the part where we want to know if is copyrightable is an
| intellectual invention of a human mind.
|
| "Ingesting someone else's code" does not seem very useful
| here - it's hardly quantifiable, nor is "ingestion" the key
| question I believe.
| Sharlin wrote:
| The point is that even a work written by an AI trained
| exclusively on liberally licensed or public domain material
| cannot have copyright (isn't a "work" in the legal sense) and
| thus nobody has standing to put it under a license or claim any
| rights to it.
|
| If I train a limerick generator on the contents of Project
| Gutenberg, no matter how creative its outputs, they're not
| copyrightable under this interpretation. And it's by far the
| most reasonable interpretation of the law as both intended and
| written. Entities that are not legal persons cannot have
| copyright, but legal persons also cannot claim copyright of
| something made by a nonperson, unless they are the "creative
| force" behind the work.
| __alexs wrote:
| AI written absolutely is copyrightable. There are just some
| unresolved tensions around where the lines are and how much and
| what kind of involvement humans need to have in the process.
| cxr wrote:
| FYI: the concept is "copyright" not "copywrite". It doesn't
| turn into "copywritten" as an adjective. The adjective is
| "copyrighted".
| zozbot234 wrote:
| If you ask a LLM to derive a spec that has no expressive element
| of the original code (a clean-room human team can carefully
| verify this), and then ask another instance of the LLM (with
| fresh context) to write out code from the spec, how is that
| different from a "clean room" rewrite? The agent that writes the
| new code only ever sees the spec, and by assumption (the
| assumption that's made in all clean room rewrites) the spec is
| purely factual with all copyrightable expression having been
| distilled out.
| miroljub wrote:
| The new agent who writes code has probably at least parts of
| the original code as training data.
|
| We can't speak about clean room implementation from LLM since
| they are technically capable only of spitting their training
| data in different ways, not of any original creation.
| nubg wrote:
| Only in the case of open source code
| dizhn wrote:
| The conclusion of this would be that you can never license AI
| generated code since you can't get a release from the
| original authors.
|
| Of course in practice it would work exactly in the opposite
| fashion and AI generated code would be immune even if it
| copied code verbatim.
| jesterswilde wrote:
| I don't see what's wrong with that personally. If I pirated
| someone's software, and then sold it as my own and got
| caught, just because I sold a bunch of it doesn't mean
| those people who bought it now are in the clear. They are
| still using bootleg software in their business.
| gf000 wrote:
| I guess it depends on if the source data set is part of the
| training data or not (if it's open source it is likely part of
| it).
|
| A lawyer could easily argue that the model itself stores a
| representation of the original, and thus it can never do a
| "fresh context".
|
| And to be perfectly honest, LLMs can quote a _lot_ of text
| verbatim.
| k__ wrote:
| How do you prove the training data didn't contain the code?
|
| I'd assume an LLM trained on the original would also be
| contaminated.
| est wrote:
| Uh, patricide?
|
| The key leap from gpt3 to gpt-3.5 (aka ChatGPT) was code-
| davinci-002, which is trained upon Github source code after
| OpenAI-Microsoft partnership.
|
| Open source code contributed much to LLM's amazing CoT
| consistency. If there's no Open Source movement, LLM would be
| developed much later.
| pu_pe wrote:
| Licensing issues aside, the chardet rewrite seems to be clearly
| superior to the original in performance too. It's likely that
| many open source projects could benefit from a similar approach.
| spwa4 wrote:
| Can we do the same with universal music? Because that's easy and
| already possible. Or Microsoft Windows? Because we all know the
| answer: if it works, essentially any government will immediately
| call it illegal.
|
| Because if this isn't allowed, that makes all of the AI models
| _themselves_ illegal. They are very much the product of using
| others ' copyrighted stuff and rewriting it.
|
| But of course this will be allowed because copyright was never
| meant to protect anyone small. And that it's in direct
| contradiction with what applies to large companies? Courts won't
| care.
| gspr wrote:
| The dark future possibility here is that the big guy is allowed
| to launder the intellectual property of the little guy, but not
| vice versa.
| vetrom wrote:
| That dark future is now, look at case law as applied to the
| AI operators vs the 'little guys'.
| spwa4 wrote:
| Even big copyright firms. Disney especially is known for
| rehashing existing material and then not allowing anyone
| else to do the same with their stuff. Disney does not have
| a lot of original stories.
| blamestross wrote:
| Intellectual property laundering is the core and primary value of
| LLMs. Everything else is "bonus".
| gspr wrote:
| > If "AI-rewriting" is accepted as a valid way to change
| licenses, it represents the end of Copyleft. Any developer could
| take a GPL-licensed project, feed it into an LLM with the prompt
| "Rewrite this in a different style," and release it under MIT.
| The legal and ethical lines are still being drawn, and the
| chardet v7.0.0 case is one of the first real-world tests.
|
| This isn't even limited to "the end of copyleft"; it's the end of
| all copyright! At least copyright protecting the little guy. If
| you have deep enough pockets to create LLMs, you can in this
| potential future use them to wash away anyone's copyright for any
| work. Why would the GPL be the only target? If it works for the
| GPL, it surely also works for your photographs, poetry - or hell
| even proprietary software?
| abrookewood wrote:
| This seems relevant: "No right to relicense this project
| (github.com/chardet)"
| https://news.ycombinator.com/item?id=47259177
| shevy-java wrote:
| That's another project though, right? In this case I think it
| is different because that project just seems stolen. The courts
| can probably verify this too.
|
| I think the main question is when a rewrite is a clean rewrite,
| via AI. If it is a clean rewrite they can choose any licence.
| littlestymaar wrote:
| No, TFA is about chardet too:
|
| > chardet , a Python character encoding detector used by
| requests and many others, has sat in that tension for years:
| as a port of Mozilla's C++ code it was bound to the LGPL,
| making it a gray area for corporate users and a headache for
| its most famous consumer.
| emsign wrote:
| By design you can't know if the LLM doing the rewrite was exposed
| to the original code base. Unless the AI company is disclosing
| their training material, which they won't because they don't want
| to admit breaking the law.
| d1sxeyes wrote:
| Is it against the law for an LLM to read LGPL-licensed code?
|
| That's a complex question that isn't solved yet. Clearly,
| regurgitating verbatim LGPL code in large chunks would be
| unlawful. What's much less clear is a) how large do those
| chunks need to be to trigger LGPL violations? A single line?
| Two? A function? What if it's trivial? And b) are all outputs
| of a system which has received LGPL code as an input
| necessarily derivative?
|
| If I learn how to code in Python exclusively from reading LGPL
| code, and then go away and write something new, it's clear that
| I haven't committed any violation of copyright under existing
| law, even if all I'm doing as a human is rearranging tokens I
| understand from reading LGPL code semantically to achieve new
| result.
|
| It's a trying time for software and the legal system. I don't
| have the answers, but whether you like them or not, these
| systems are here to stay, and we need to learn how to live with
| them.
| gostsamo wrote:
| it was exposed when it was shown the thing to rewrite.
| shevy-java wrote:
| In this context here I think that is a correct statement. But
| I think you can have LLMs that can generate the same or
| similar code, without having been exposed to the other code.
| skeledrew wrote:
| It doesn't even matter if the LLM was exposed during training.
| A clean-room rewrite can be done by having one LLM create a
| highly detailed analysis of the target (reverse engineering if
| it's in binary form), and providing that analysis to another
| LLM to base an implementation.
| xyzsparetimexyz wrote:
| Why does it need 2 LLMs? LLMs aren't people. I'm not even
| sure that it needs to be done in 2 seperate contexts
| shevy-java wrote:
| Agreed. But even then I don't see the problem. Multiple
| LLMs could work on the same project.
| skeledrew wrote:
| It doesn't have to be 2 LLMs, but nowadays there's LLM
| auto-memory, which means it could be argued that the same
| LLM doing both analysis and reimplementation isn't "clean".
| And the entire purpose behind the "clean" is to avoid that
| argument.
| k__ wrote:
| It doesn't matter for the LLM writing the analysis.
|
| It does matter for the one who implements it.
|
| Finding an LLM that's good enough to do the rewrite while
| being able to prove it wasn't exposed to the original GPL
| code is probably impossible.
| soulofmischief wrote:
| Seeing the source for a project doesn't prevent me from ever
| creating a similar project, just because I've seen the code.
| The devil is in the details.
| shevy-java wrote:
| Agreed, but the courts can conclude that all LLMs who are not
| open about their decision, have stolen things. So LLMs would
| auto-lose in court.
| orthoxerox wrote:
| Or they can conclude otherwise.
| shevy-java wrote:
| > By design you can't know if the LLM doing the rewrite was
| exposed to the original code base.
|
| I agree, in theory. In practice courts will request that the
| decision-making process will be made public. The "we don't
| know" excuse won't hold; real people also need to tell the
| truth in court. LLMs may not lie to the court or use the
| chewbacca defence.
|
| Also, I am pretty certain you CAN have AI models that explain
| how they originated to the decision-making process. And they
| can generate valid code too, so anything can be autogenerated
| here - in theory.
| airforce1 wrote:
| I don't see how this is different from current human poaching
| practices. i.e. It appears to be currently legal to hire an
| employee from company A who has been "tainted" by company A's
| [proprietary AI secrets/proprietary CPU architecture
| secrets/etc] in order to develop a competing offering for
| company B. i.e. It's not illegal for a human who worked at
| Intel for 20 years to go work for AMD even though they are
| certainly "tainted" with all sorts of copyrighted/proprietary
| knowledge that will surely leak through at AMD. Maybe patents
| are a first line of defense for company A, but that can't
| prevent adjacent solutions that aren't outright duplications
| and circumvent the patent.
| gbuk2013 wrote:
| In mind, if you feed code into an AI model then the output is
| clearly a derivative work, with all the licensing implications.
| This seems objectively reasonable?
| quotemstr wrote:
| Nobody in this discussion knows what the words "derivative" and
| "work" mean individually, much less together
| MagicMoonlight wrote:
| But code was fed into the models to create them, so where do we
| draw the line?
| gbuk2013 wrote:
| That is a good question - my personal opinion is that it
| should mean that models are not subject to copyright at all
| (similar to databases) but we will see what the courts decide
| :)
| Cantinflas wrote:
| "If "AI-rewriting" is accepted as a valid way to change licenses,
| it represents the end of Copyleft. "
|
| Software in the AI era is not that important.
|
| Copyleft has already won, you can have new code in 40 seconds for
| $0.70 worth of tokens.
| p0w3n3d wrote:
| Just take the code and let it AI rewrite. But... AI was taught
| on all the OpenSource Code available. Lot of them were GPL I
| think... So...
| WesolyKubeczek wrote:
| Let's then abolish all copyright on all software, what ever
| could go wrong?
| Cantinflas wrote:
| It doesn't matter any more! Software is a (very cheap)
| commodity now.
| skeledrew wrote:
| Looks like copyright just died.
| cedws wrote:
| *for ordinary people. If you use AI to steal from rich and
| powerful people, expect the law to come down on you like a
| tonne of bricks. If you steal from authors, artists, and
| developers no worries.
| shevy-java wrote:
| > In traditional software law, a "clean room" rewrite requires
| two teams
|
| So, I dislike AI and wish it would disappear, BUT!
|
| The argument is strange here, because ... how can a2mark ensure
| that AI did NOT do a clean-room conforming rewrite? Because I
| think in theory AI can do precisely this; you just need to make
| sure that the model used does that too. And this can be verified,
| in theory. So I don't fully understand a2mark here. Yes, AI may
| make use of the original source code, but it could "implement"
| things on its own. Ultimately this is finite complexity, not
| infinite complexity. I think a2mark's argument is in theory weak
| here. And I say this as someone who dislikes AI. The main
| question is: can computers do a clean rewrite, in principle? And
| I think the answer is yes. That is not saying that claude did
| this here, mind you; I really don't know the particulars. But the
| underlying principle? I don't see why AI could not do this.
| a2mark may need to reconsider the statement here.
| titanomachy wrote:
| The foundation model probably includes the original project in
| its training set, which might be enough for a court to consider
| it "contaminated". Training a new foundation model without it
| is technically possible, but would take months and cost
| millions of dollars.
| __alexs wrote:
| I think the problem here is that an AI is not a legal entity.
| It doesn't matter if you as individual run an AI that takes the
| source, dumps out a spec that you then feed into another AI.
| The legal liability lies with the operator of the AI, the
| original copyleft license was granted to a person, not to a
| robot.
|
| Now if you had 2 entirely distinct humans involved in the
| process that might work though.
| dspillett wrote:
| _> how can a2mark ensure that AI did NOT do a clean-room
| conforming rewrite?_
|
| In cases like this it is usually incumbent on the entity
| claiming the clean-room situation was pure to show their
| working. For instance how Compaq clean-room cloned the IBM BIOS
| chip1 was well documented (the procedures used, records of
| comms by the teams involved) where some other manufacturers did
| face costly legal troubles from IBM.
|
| So the question is "is the clean-room claim sufficiently backed
| up to stand legal tests?" [and moral tests, though the AI world
| generally doesn't care about failing those]
|
| --------
|
| [1] the one part of their PCs that was not essentially off-the-
| shelf, so once it could be reliably legally mimicked this
| created an open IBM PC clone market
| orthoxerox wrote:
| Clean room is sufficient, but not necessary to avoid the
| accusations of license violation.
|
| a2mark has to demonstrate that v7 is "a work containing the v6
| or a portion of it, either verbatim or with modifications
| and/or translated straightforwardly into another language",
| which is different from demanding a clean-room
| reimplementation.
|
| Theoretically, the existence of a publicly available commit
| that is half v6 code and half v7 can be used to show that this
| part of v7 code has been infected by LGPL and must thus infect
| the rest of v7, but that's IMO going against the spirit of the
| [L]GPL.
| Orygin wrote:
| Please don't use loaded terms like "infect". The license does
| not infect, it has provisions and requirements. If you want
| to interact with it, you either accept them or don't use the
| project. In this case, the author of v7 is trying to steal
| the copyrighted work of other authors by re-licensing it
| illegally.
| orthoxerox wrote:
| Is their work present in v7?
| Orygin wrote:
| Yes. Commits clearly show in progress where both LGPL and
| MIT code was working together. This clearly show they are
| a derivative work and MUST follow the original license.
|
| Plus the argument put forth is that they can _re-license_
| the project. It 's not a new one made from scratch.
| orthoxerox wrote:
| So, if these commits were private and squashed together
| before 7.0 was published there would be no violation?
| duskdozer wrote:
| They would be concealing the violation.
| orthoxerox wrote:
| Consider TCC relicensing. They identified the files
| touched by contributors that wanted to keep the GPL
| license and reimplemented them. No team A/team B clean
| room approach used. The same happened here, but at a
| different scale. All files now have a new author and this
| author is free to change the license of his work.
| Orygin wrote:
| The commits being public or not does not change the fact
| the developement was made as a derivative work of the
| original version.
| tzs wrote:
| Did they eventually remove/replace all the LGPL code?
| duskdozer wrote:
| Yes. The AI operator posted this as the prompt: https://g
| ithub.com/chardet/chardet/commit/f51f523506a73f89f0...
|
| which, minimally instructs it to directly examine the
| test suite: `4. High encoding accuracy on the chardet
| test suite`
| orthoxerox wrote:
| So what? Is reading code the same as copying code or
| modifying existing code?
| Orygin wrote:
| If you want to prove you did not make a derivative work,
| yes it helps if you never read the source code. Hence so
| call "clean room" implementations.
| orthoxerox wrote:
| Why should I prove that? Let those who claim the
| violation prove that.
| Orygin wrote:
| There is plenty of evidence already. The claim has been
| substantiated.
|
| You can't just dismiss it then say the claimant has to
| provide proof.
| foltik wrote:
| Turns out there's no need to speculate. Someone pointed out on
| GH [0] that the AI was literally prompted to copy the existing
| code:
|
| > *Context:* The registry maps every supported encoding to its
| metadata. Era assignments MUST match chardet 6.0.0's
| `chardet/metadata/charsets.py` at
| https://raw.githubusercontent.com/chardet/chardet/f0676c0d6a...
|
| > Fetch that file and use it as the authoritative reference for
| which encodings belong to which era. Do not invent era
| assignments.
|
| [0]
| https://github.com/chardet/chardet/issues/327#issuecomment-4...
| orthoxerox wrote:
| That's data, not code.
| foltik wrote:
| It's a python file from chardet 6, doesn't matter what you
| think it does. It clearly wasn't a clean room
| reimplementation.
| duskdozer wrote:
| This is such scummy behavior.
| tgma wrote:
| Isn't AFC test applicable here?
| amelius wrote:
| I think you should interpret it like this:
|
| You cannot copyright the alphabet, but you can copyright the way
| letters are put together.
|
| Now, with AI the abstraction level goes from individual letters
| to functions, classes, and maybe even entire files.
|
| You can't copyright those (when written using AI), but you
| __can__ copyright the way they are put together.
| josephg wrote:
| > You can't copyright those anymore (when written using AI),
| but you __can__ copyright the way they are put together.
|
| Sort of, but not really. Copyright usually applies to a
| _specific work_. You can copyright Harry Potter. But you can 't
| copyright the general class of "Wizard boy goes to wizard
| school". Copyrights generally can't be applied to classes of
| works. Only one specific work. (Direct copies - eg made with a
| photocopier - are still considered the same work.)
|
| Patterns (of all sorts) usually fall under patent law, not
| copyright law. Patents have some additional requirements -
| notably including that a patent must be novel and non-obvious.
| I broadly think software patents are a bad idea. Software is
| usually obvious. Patents stifle innovation.
|
| Is an AI "copy" a copy like a photocopier would make? Or is it
| a novel work? It seems more like the latter to me. An AI copy
| of a program (via a spec) won't be a copy of the original code.
| It'll be programmed differently. Thats why "clean room
| reimplementations" are a thing - because doing that process
| means you _can 't_ just copy the code itself. But what do I
| know, I'm not a lawyer or a judge. I think we'll have to wait
| for this stuff to shake out before anyone really knows what the
| rules will end up being.
|
| Weird variants of a lot of this stuff have been tested in
| court. Eg the Google v Oracle case from a few years ago.
| amelius wrote:
| You have good points regarding how copyright works.
|
| > Software is usually obvious.
|
| Hardware and mechanical designs are usually described in CAD
| programs nowadays, so it comes pretty close to software; it's
| just that LLMs are not the right tool to "GenAI" them but
| I've seen plenty of these kinds of design that I know for
| sure that they are often not any less obvious than a lot of
| software. Treating software as "obvious therefore not
| patentable" is not accurate and not fair and is probably not
| going to help the profession in the AI age. But I agree that
| patents are bad for innovation.
|
| It is also not fair to claim that an AI-copy is fundamentally
| different from photocopying.
|
| I mean, in both cases it is like you are picking the worst
| case interpretation for the field of software engineering.
|
| > I think we'll have to wait for this stuff to shake out
| before anyone really knows what the rules will end up being.
|
| Yes, but it will help if we think deeply about this stuff
| ourselves because what law-makers come up with may not be
| what the profession needs.
| josephg wrote:
| > It is also not fair to claim that an AI-copy is
| fundamentally different from photocopying.
|
| If you clean-room copy it, I think it is different. Eg,
| first get one agent to make a complete spec of what the
| program does. And a list of all the correctness guarantees
| it meets. Then feed that spec into another AI model to
| generate a program which meets that spec.
|
| The second program will not be based on any of the code in
| the first program. They'll be as different as any two
| implementations of the same idea are. I don't think the
| second program should be copyrighted. If it should, why
| shouldn't one C compiler should be able to own a copyright
| over all C compilers? Why doesn't the first JSON parsing
| library own JSON parsing? These seem the same to me. I
| don't see how AI models change anything, other than taking
| human effort out of the porting process.
| amelius wrote:
| The output will still be dependent on the input. And it
| is still copying even if you first lift the input to a
| different abstraction level.
|
| Finally, even if your rationale is 99% correct, then
| there is still that 1% that makes the result a
| mechanistic copy.
|
| And I see no way in which most people would 100% agree
| with your view.
| josephg wrote:
| If you write program A that does something, and I look at
| what your program does and write program B that does the
| same thing, have I copied your program? So long as I
| didn't copy any of the lines of code in program A
| directly, no. At least, not according to copyright law. A
| copyright on Netscape navigator doesn't apply to internet
| explorer or chrome. They're all "copies" of Netscape
| navigator. But copyright applies to the work. New work?
| New copyright. I really don't see how an LLM being
| involved changes any of that.
|
| If you want to protect the idea or the design, get a
| patent. A patent on one h264 encoder applies to all h264
| encoders.
|
| There is a chance the courts or the legislature will
| decide differently. But until then, we should assume the
| existing law of the land holds.
| dessimus wrote:
| Interesting to see how this plays out. Conceivably if running an
| LLM over text defeats copyright, it will destroy the book
| publishing industry, as I could run any ebook thru an LLM to make
| a new text, like the ~95% regurgitated Harry Potter.
| amelius wrote:
| If enough people do this, then it may speed up the lawmaking
| process.
| timschmidt wrote:
| This has already been done via brute force for melodies:
| https://www.vice.com/en/article/musicians-algorithmically-ge...
| amelius wrote:
| Did they listen to their own creation?
|
| If not, maybe it should not constitute a valid case in court.
|
| Also, I'm wondering if they are not themselves liable
| considering they have every copyrighted work in there too.
| kingstnap wrote:
| You could already do that before LLMs?
|
| Persumably there is already a law around why I cant just go
| borrow a book from my library, type out some 95% regurgitated
| varient on my laptop, and then try to publish it somewhere?
|
| Edit: I looked it up and the thing that stops you from
| publishing a bootleg "Harold Potter and the Wizards Rock" is
| this legal framework around "The Abstractions Test".
| dessimus wrote:
| I agree, but I'm not the one claiming that an AI-Assisted
| rewrite is sufficient enough to now claim that one ignore
| copyright and change the license.
| dspillett wrote:
| _> Accepting AI-rewriting as relicensing could spell the end of
| Copyleft_
|
| The more restrictive licences perhaps, though only if the
| rewriter convinces everyone that they can properly maintain the
| result. For ancient projects that aren't actively maintained
| anyway (because they are essentially _done_ at this point) this
| might make little difference, but for active projects any new
| features and fixes might result in either manual reimplementation
| in the rewritten version or the clean-room process being repeated
| completely for the whole project.
|
| _> chardet 7.0 is a ground-up, MIT-licensed rewrite of chardet.
| Same package name, same public API -- _
|
| (from the github description)
|
| The "same name" part to me feels somewhat disingenuous. It _isn
| 't_ the same thing so it should have a different name to avoid
| confusion, even if that name is something very similar to the
| original like chardet-ng or chardet-ai.
| conartist6 wrote:
| Who cares if it can be maintained. The system now penalizes the
| original creator for creating it and gives thieves the ability
| to conduct legal theft at a gargantuan scale, the only limit
| being how creative the abuser is in making money.
|
| With the incentives set up like that, the era of open software
| cooperation would be ended rapidly.
| dspillett wrote:
| _> Who cares if it can be maintained._
|
| People who understand and care about the implications of
| https://xkcd.com/2347/
|
| Which admittedly is not nearly enough of us...
| throwayaw84330 wrote:
| This is super interesting. Exploring the basis for Free
| Software (the 4 liberties, Richard Stallman)... if AI-code is
| effectively under Public Domain, wouldn't that actually be even
| MORE defensive than relying on copyright to be able to generate
| copyleft? Wouldn't the rewrite of code (previously under any
| license, and maybe even unknown to the LLM) constitute a
| massive win for the population in general, because now their 4
| liberties are more attainable through the extensive use of LLMs
| to generate code?
| dspillett wrote:
| Many copyleft licences give more rights to the user of the
| software than being public domain would.
|
| A bit of public domain code can be used in a hidden way in
| perpetuity.
|
| A bit of code covered by AGPL3 (for instance) (and other GPLs
| depending on context) can be used for free too, but with the
| extra requirement that users be given a copy of the code, and
| derivative works, upon request.
|
| This is why the corps like MIT and similar and won't touch
| anything remotely like GPL (even LGPL which only covers
| derivative works of the library not the wider project). The
| MIT licence can be largely treated as public domain.
| RcouF1uZ4gsC wrote:
| > The copyright vacuum: If AI-generated code cannot be
| copyrighted (as the courts suggest), then the maintainers may not
| even have the legal standing to license v7.0.0 under MIT or any
| license.
|
| I believe this is a misunderstanding of the ruling. The code
| can't be copyrighted by a LLM. However, the code could be
| copyrighted by the person running the LLM.
| stuaxo wrote:
| I don't see how (with current LLMs that have been trained on
| mixed licensed data) you can use the LLM to rewrite to a less
| restrictive license.
|
| You could probably use it to output code that is GPL'd though.
| danlitt wrote:
| I am pretty sure this article is predicated on a misunderstanding
| of what a "clean room" implementation means. It does not mean "as
| long as you never read the original code, whatever you write is
| yours". If you had a hermetically sealed code base that _just
| happened_ to coincide line for line with the codebase for GCC, it
| would still be a copy. Traditionally, a human-driven clean room
| implementation would have a vanishingly small probability of
| matching the original codebase enough to be considered a copy.
| With LLMs, the probability is much higher (since in truth they
| are very much not a "clean room" at all).
|
| The actual meaning of a "clean room implementation" is that it is
| derived from an API and not from an implementation (I am
| simplifying slightly). Whether the reimplementation is actually a
| "new implementation" is a subjective but empirical question that
| basically hinges on how similar the new codebase is to the old
| one. If it's too similar, it's a copy.
|
| What the chardet maintainers have done here is legally very
| irresponsible. There is no easy way to guarantee that their code
| is actually MIT and not LGPL without auditing the entire
| codebase. Any downstream user of the library is at risk of the
| license switching from underneath them. Ideally, this would burn
| their reputation as responsible maintainers, and result in
| someone else taking over the project. In reality, probably it
| will remain MIT for a couple of years and then suddenly there
| will be a "supply chain issue" like there was for mimemagic a few
| years ago.
| dathinab wrote:
| the author speaks about code which is syntactically completely
| different but semantically does the same
|
| i.e. a re-implementation
|
| which can either
|
| - be still derived work, i.e. seen as you just obfuscating a
| copyright violation
|
| - be a new work doing the same
|
| nothing prevents an AI from producing a spec based on a API,
| API documentation and API usage/fuzzing and then resetting the
| AI and using that spec to produce a rewrite
|
| I mean "doing the same" is NOT copyright protection, you need
| patent law for that. Except even with patent law you need to
| innovations/concepts not the exact implementation details.
| Which means that even if there are software patents
| (theoretically,1) most things done in software wouldn't be
| patentable (as they are just implementation details, not
| inventions)
|
| (1): I say theoretically because there is a very long track
| record of a lot of patents being granted which really should
| never be granted. This combined with the high cost of
| invalidating patents has caused a ton of economical damages.
| jacquesm wrote:
| No, that depends on whether or not the AI work product rests
| on key contributions to its training set without which it
| would not be able to the the work, see other comment. In that
| case it looks like 'a new work doing the same' but it still a
| derived work.
|
| Ted Nelson was years ahead of the future where we really
| needed his Xanadu to keep track of fractional copyright.
| Likely if we had such a mechanism, and AI authors respected
| it then we would be able to say that your work is derived
| from 3000 other original works and that you added 6 lines of
| new code.
| uyzstvqs wrote:
| No, training and inference are two separate processes.
| Training data is never redistributed, only obtained and
| analyzed. What matters is what data is put into _context_
| during inference. This is controlled by the user.
|
| AI/ML is complex, so as a simpler analogy: If I watch The
| Simpsons, and I create an amusing infographic of how often
| Homer says "D'oh!" over time, my infographic would be an
| original work. AI training follows the same principle.
| jacquesm wrote:
| > my infographic would be an original work.
|
| > AI training follows the same principle.
|
| If you really believe that then we can't have a
| meaningful conversation about this, that's not even ELIF
| territory, that's just disconnected. You should be asking
| questions, not telling people how it works.
| ndriscoll wrote:
| How exactly is it different? All the model itself is is a
| probability distribution for next token given input,
| fitted to a giant corpus. i.e. a description of
| statistical properties. On its own it doesn't even "do"
| anything, but even if you wrap that in a text generator
| and feed it literal gcc source code fragments as input
| context, it will quickly diverge. Because it's not a copy
| of gcc. It doesn't contain a copy of gcc. It's a
| description of what language is common in code in
| general.
|
| In fact we could make this concrete: use the model as the
| prediction stage in a compressor, and compress gcc with
| it. The residual is the extent to which it doesn't
| contain gcc.
| jacquesm wrote:
| There already have been multiple documented cases of LLMs
| spitting out fairly large chunks of the input corpus.
| There have been some experiments to get it to replicate
| the entirety of 'Moby Dick' with some success for one
| model but less success with others most likely due to
| output filtering to prevent the generation of such texts,
| but that doesn't mean they're not in there in some form.
| And how could they not be, it is just a lossy compression
| mechanism, the degree of loss is not really all that
| relevant to the discussion.
| ndriscoll wrote:
| Are you referring to this?
|
| https://osyuksel.github.io/blog/reconstructing-moby-dick-
| llm...
|
| I see a test where one model managed to 85% reproduce a
| _paragraph_ given 3 input paragraphs under 50% of the
| time.
|
| So it can't even produce 1 paragraph given 3 as input,
| and it can't even get close half the time.
|
| "Contains Moby Dick" would be something like you give it
| the first paragraph and it produces the rest of the book.
| What we have here instead is a statistical model that
| when given passages can do an okay job at predicting a
| sentence or two, but otherwise quickly diverges.
| xyzzy_plugh wrote:
| I'm no longer certain what point you're trying to make.
|
| Getting close less than half the time given three
| paragraphs as input still sounds like red-handed
| copyright infringement to me.
|
| If I sample a copyrighted song in my new track, clip it,
| slow it down, and decimate the bit rate, a court would
| not let me off the hook.
|
| It doesn't matter how much context you push into these
| things. If I feed them 50% of Moby Dick and they produce
| the next _word_ , and I can repeatedly do that to produce
| the entire book (I'm pretty sure the number of attempts
| is wholly irrelevant: we're impossibly far from monkeys
| on typewriters) then we can prove the statistical model
| encodes the book. The further we are from that (and the
| more we can generate with less) then the stronger the
| case is. It's a pretty strong case!
| ndriscoll wrote:
| That's... not how this works.
|
| > If I feed them 50% of Moby Dick and they produce the
| next _word_ and I can repeatedly do that to produce the
| entire book... then we can prove the statistical model
| encodes the book.
|
| It can't because it doesn't. That's what it means to say
| it diverges.
|
| The "number of attempts" is you cheating. _You 're giving
| it the book when you let it try again word by word until
| it gets the correct answer_, and then claiming it
| produced the book. That's exactly the residual that I
| said characterizes the extent to which it doesn't know
| the book. Trivially, no matter how bad the model is, if
| you give it the residual, it can losslessly compress
| anything at all.
|
| If you had a simple model that just predicts next word
| given current word (trained on word pair frequency across
| all English text, or even all text excluding Moby Dick),
| and then give it retries until it gets the current word
| right, it will also quickly produce the book. Because it
| was your retry policy that encoded the book, not the
| model. Without that policy, it will get it wrong within a
| few words, just like these models do.
| jacquesm wrote:
| The point is that the AI's themselves _and_ their backers
| are on the record as saying that the AI _could_ reproduce
| copyrighted works in their entirety but that there are
| countermeasures in place to stop them from doing so.
|
| I wonder what the results would be if I spent time to
| train a model up from scratch without any such
| constraints. But I'm much too busy with other stuff right
| now, but that would be an interesting challenge.
| ndriscoll wrote:
| Yeah just like a star _could_ appear inside of Earth from
| quantum pair production at any given moment. But
| realistically, it can 't. And you can't even show a test
| where any model can get more than a few tokens in a row
| correct.
|
| These companies just don't want to deal with people
| complaining that it reproduces something when they don't
| understand that they're literally giving it the answer.
| jacquesm wrote:
| You do realize you are now arguing against your own case
| don't you?
| xyzzy_plugh wrote:
| But it _does_ encode it! Each subsequent token 's
| probability space encodes the next word(s) of the book
| with a non-zero probability that is significantly higher
| than random noise.
|
| If you had access to a model's top p selection then I'd
| bet the book is in there consistently for every token. Is
| it statistically significant? Might be!
|
| I'm not cheating because the number of attempts is _so
| low_ it 's irrelevant.
|
| If I were to take a copyrighted work and chunk it up into
| 1000 pieces and encrypt each piece with a unique key, and
| give you all the pieces and keys, would it still be the
| copyrighted work? What if I shave off the last bit of
| each key before I give them to you, so you have a 50%
| chance of guessing the correct key for each piece? What
| if I shave two bits? What if it's a million pieces? When
| does it become transformative or no longer infringing for
| me to distribute?
|
| The answer might surprise you.
| ndriscoll wrote:
| Your test is more like the following:
|
| Consider a password consisting of random words each
| chosen from a 4k dictionary. Say you choose 10 words.
| Then your password has log_2(4k)*10 entropy.
|
| Now consider a validator that tells you when you gets a
| word right. Then you can guess one word at a time, and
| your password strength is log_2(4k*10). Exponentially
| weaker.
|
| You're constructing the second scenario and pretending
| it's the first.
|
| Also in your 50% probability scenario, each word is 1
| bit, and even 50-100 bits is unguessable. A 1000 word key
| where each word provides 1 bit would be absurdly strong.
| xyzzy_plugh wrote:
| You're still missing the point. The numbers don't matter
| because it's copyright infringement as long as I can get
| the book out. As long as I know the key, or the seed, I
| can get the book out. In court, how would you prove it's
| not infringement?
| ndriscoll wrote:
| Because _you_ put the book in. Again, this is measurable.
| Compress the book with a model as the predictor. The
| residual is you having to give it the answer. It 's
| literally you telling it the book.
| gus_massa wrote:
| For an infographic, perhaps you claim claim fair use. I
| think it makes a lot of sense, but IANAL.
|
| For a fan fiction episode that is different from all
| official episodes, you may cross your fingers.
|
| For a remake of one of the episodes with a different
| camera angle and similar dialog, I expect that you will
| get in problems.
| ndriscoll wrote:
| Is the claim that these models can 1 shot a Simpsons
| episode remake with different camera angle and similar
| dialog from a prompt like "produce Simpsons episode
| S01E04"? Or are we falling into the "the user doesn't
| notice that they told the model the answer, and the model
| in fact did not memorize the thing" trap?
| pmarreck wrote:
| > With LLMs, the probability is much higher (since in truth
| they are very much not a "clean room" at all).
|
| I beg to differ. Please examine any of my recent codebases on
| github (same username); I have cleanroom-reimplemented par2
| (par2z), bzip2 (bzip2z), rar (rarz), 7zip (z7z), so maybe I am
| a good test case for this (I haven't announced this anywhere
| until now, right here, so here we go...)
|
| https://github.com/pmarreck?tab=repositories&type=source
|
| I was most particular about the 7zip reimplementation since it
| is the most likely to be contentious. Here is my repo with the
| full spec that was created by the "dirty team" and then worked
| off of by the LLM with zero access to the original source:
| https://github.com/pmarreck/7z-cleanroom-spec
|
| Not only are they rewritten in a completely different language,
| but to my knowledge they are also completely different
| semantically _except where they cannot be to comply with the
| specification_. I invite you and anyone else to compare them to
| the original source and find overt similarities.
|
| With all of these, I included two-way interoperation tests with
| the original tooling to ensure compatibility with the spec.
| airza wrote:
| By what means did you make sure your LLM was not trained with
| data from the original source code?
| MrManatee wrote:
| Exactly - it very likely was trained on it. I tried this
| with Opus 4.6. I turned off web searches and other tool
| calls, and asked it to list some filenames it remembers
| being in the 7-zip repo. It got dozens exactly right and
| only two incorrect (they were close but not exact matches).
| I then asked it to give me the source code of a function I
| picked randomly, and it got the signature spot on, but not
| the contents.
|
| My understanding of cleanroom is that the person/team
| programming is supposed to have never seen any of the
| original code. The agent is more like someone who has read
| the original code line by line, but doesn't remember all
| the details - and isn't allowed to check.
| pmarreck wrote:
| Because it's written in an entirely different language,
| which makes this whole point moot
| sobjornstad wrote:
| Surely if I took a program written in Python and
| translated it line for line into JavaScript, that
| wouldn't allow me to treat it as original work. I don't
| see how this solves the problem, except very
| incrementally.
| ostacke wrote:
| Bu that's not really what danlitt said, right? They did not
| claim that it's impossible for an LLM to generate something
| different, merely that it's not a clean room implementation
| since the LLM, one must assume, is trained on the code it's
| re-implementing.
| galaxyLogic wrote:
| BUt LLM has seen millions (?) of other code-bases too. If
| you give it a functional spec it has no reason to prefer
| any one of those code-bases in particular. Except perhaps
| if it has seen the original spec (if such can be read from
| public sources) associated with the old implementation, and
| the new spec is a copy of the old spec.
| sarchertech wrote:
| Yes if you are solving the exact problem that the
| original code solved and that original code was labeled
| as solving that exact problem then that's very good
| reason for the LLM to produce that code.
|
| Researchers have shown that an LLM was able to reproduce
| the verbatim text of the first 4 Harry Potter books with
| 96% accuracy.
| pmarreck wrote:
| This is not an argument against coding in a different
| language, though. It would be like having it restate
| Harry Potter in a different language with different main
| character names, and reshuffled plot points.
| 0x457 wrote:
| > that an LLM was able to reproduce the verbatim text of
| the first 4 Harry Potter books with 96% accuracy.
|
| Kinda weird argument, in their research
| (https://forum.gnoppix.org/t/researchers-extract-up-
| to-96-of-...) LLM was explicitly asked to reproduce the
| book. There are people that can do so without LLMs out
| there, by this logic everything they write is a copyright
| infringement an every book they can reproduce.
|
| > Yes if you are solving the exact problem that the
| original code solved and that original code was labeled
| as solving that exact problem then that's very good
| reason for the LLM to produce that code.
|
| I think you're overestimating LLM ability to generalize.
| pmarreck wrote:
| Well, if you're coding it in Zig, and it's barely seen any
| Zig, then how exactly would that argument hold up in that
| case?
| danlitt wrote:
| I only said the probability is _higher_ , not that the
| probability is 1!
| zabzonk wrote:
| > It does not mean "as long as you never read the original
| code, whatever you write is yours"
|
| I think there is precedence that says exactly this - for
| example the BIOS rewrites for the IBM PC from people like
| Phoenix. And it would be trivial to instruct an LLM to prefer
| to use (say, in assembler) register C over register B wherever
| that was possible, resulting in different code.
| bandrami wrote:
| Different but still derivative
| zabzonk wrote:
| Well, I am not exactly a hotshot 8086 programmer (though I
| do alright) but if I was asked to reproduce the IBM BIOS
| (which I have seen) I think I would come up with something
| very similar but not identical - it is really not rocket
| science code, so the LLM replacing me would have rather few
| alternatives to choose from.
| fc417fc802 wrote:
| I believe those are actually separate matters. A proper
| clean room implementation on the one hand, and the
| question of whether or not a particular outcome was a
| foregone conclusion on the other. I don't recall where I
| saw the latter but it might have come up during Google v
| Oracle?
| danlitt wrote:
| As long as you never read the original code, it _is_ very
| likely that whatever you write is yours. So I would not be
| surprised to read judges indicating in this direction. But I
| would be a little surprised to find out this was an actual
| part of the test, rather than an indication that the work was
| considered to have been copied. There are for instance lots
| of ways of reproducing copyrighted work without using a copy
| directly, but naive methods like generating random pieces of
| text are very time consuming, so there is not much precedence
| around them. LLMs are much more efficient at it!
| petercooper wrote:
| _The actual meaning of a "clean room implementation" is that it
| is derived from an API and not from an implementation_
|
| I know you were simplifying, and not to take away from your
| well-made broader point, but an API-derived implementation can
| still result in problems, as in Google vs Oracle [1]. The
| Supreme Court found in favor of Google (6-2) along "fair use"
| lines, but the case dodged setting any precedent on the nature
| of API copyrightability. I'm unaware if future cases have set
| any precedent yet, but it just came to mind.
|
| [1]: https://en.wikipedia.org/wiki/Google_LLC_v._Oracle_America
| ,_....
| lokar wrote:
| Yeah, a cleanroom re-write, or even "just" a copy of the API
| spec is something to raise as a defense during a trial (along
| with all other evidence), it's not a categorical exemption
| from the law.
|
| Also, I find it important that here the API is really minimal
| (compared to the Java std lib), the real value of the library
| is in the internal detection logic.
| danlitt wrote:
| This is exactly what I had in mind when I said I was
| simplifying :) it is a valid point.
| femto wrote:
| > If you had a hermetically sealed code base that just happened
| to coincide line for line with the codebase for GCC, it would
| still be a copy.
|
| That's not what the law says [1]. If two people happen to
| _independently_ create the same thing they each have their own
| copyright.
|
| If it's highly improbable that two works are independent (eg.
| the gcc code base), the first author would probably go to court
| claiming copying, but their case would still fail if the second
| author could show that their work was independent, no matter
| how improbable.
|
| [1]
| https://lawhandbook.sa.gov.au/ch11s13.php?lscsa_prod%5Bpage%...
| jerf wrote:
| It is true that if two people happen to independently create
| the same thing, they each have their own copyright.
|
| It is also true that in all the cases that I know about where
| that has occurred the courts have taken a very, very, _very_
| close look at the situation and taken extensive evidence to
| convince the court that there really wasn 't any copying. It
| was anything but a "get out of jail free" card; it in fact
| was difficult and expensive, in proportion to the size of the
| works under question, to prove to the court's satisfaction
| that the two things really were independent. Moreover, in all
| the cases I know about, they weren't actually _identical_ ,
| just, really really close.
|
| No rational court could possibly ever come to that conclusion
| if someone claimed a line-by-line copy of gcc was written by
| them, they must have independently come up with it. The
| probably of that is one out of ten to the "doesn't even
| remotely fit in this universe so forget about it". The bar to
| overcoming that is simply impossibly high, unlike two songs
| that happen to have similar harmonies and melodies, given the
| exponentially more constrained space of "simple song" as
| compared to a compiler suite.
| wareya wrote:
| That's the "but their case would still fail if the second
| author could show that their work was independent, no
| matter how improbable" part of the post you're responding
| to.
| jerf wrote:
| One out of ten to the power of "forget about it" is not
| improbable, it's impossible.
|
| I know it's a popular misconception that "impossible" = a
| strict, statistical, mathematical 0, but if you try to
| use that in real life it turns out to be pretty useless.
| It also tends to bother people that there isn't a bright
| shining line between "possible" and "impossible" like
| there is between "0 and strictly not 0", but all you can
| really do is deal with it. Where ever the line is, this
| is literally millions of orders of magnitude on the wrong
| side of it. Not a factor of millions, a factor of _ten to
| the_ millions. It 's not possible to "accidentally"
| duplicate a work of that size.
| wareya wrote:
| It sounds to me like you're responding to a different
| argument than they're actually making and reading intent
| into it that isn't written into it.
| gruez wrote:
| All of this is moot for the purposes of LLM, because it's
| almost certain that the LLMs were trained on the code base,
| and therefore is "tainted". You can't do this with humans
| either. Clean room design requires separate people for the
| spec/implementation.
| danlitt wrote:
| Thank you for providing a reference! I certainly admit that
| "very similar photographs are not copies" as the reference
| states. And certainly physical copying qualifies as copying
| in the sense of copyright. However I still think copying can
| happen even if you never have access to a copy.
|
| I suppose a different way of stating my position is that some
| activities that don't _look like_ copying are in fact
| copying. For instance it would not be required to find a
| literal copy of the GCC codebase inside of the LLM somehow,
| in order for the produced work to be a copy. Likewise if I
| specify that "Harry Potter and the Philosopher's Stone is
| the text file with hash
| 165hdm655g7wps576n3mra3880v2yzc5hh5cif1x9mckm2xaf5g4" and
| then someone else uses a computer to brute force find a hash
| collision, I suspect this would still be considered a copy.
|
| I think there is a substantial risk that the automatic
| translation done in this case is, at least in part, copying
| in the above sense.
| femto wrote:
| I fully agree with you. (A small information theory nit
| pick with your example. The hash and program would have to
| be at least as long as a perfectly compressed copy of Harry
| Potter and the Philosopher's Stone. If not you've just
| invented a better compressor and are in the running for a
| Hutter Prize[1]! A hash and "decomporessor" of the required
| length would likely be considered to embody the work.)
|
| It's an interesting case. As I understand it, there is an
| ongoing debate within the AI research community as to
| whether neural nets are encoding verbatim blocks of
| information or creating a model which captures the
| "essence" or "ideas" behind a work. If they are capturing
| ideas, which are not copyrightable, it would suggest that
| LLMs can be used to "launder" copyright. In this case, I
| get the feeling that, for clarity, we would both say that
| the work in question (or works derived from it) should not
| be part of the training set or prompt, emulating a clean
| room implementation by a human. (Is that a fair comment?)
|
| I've no direct experience here, but I would come down on
| the side of "LLMs are encoding (copyrightable) verbatim
| text", because others are reporting that LLMs do
| regurgitate word-for-word chunks of text. Is this always
| the case though? Do different AI architectures, or models
| that are less well fitted, encode ideas rather than quotes?
|
| [1] https://en.wikipedia.org/wiki/Hutter_Prize
|
| Edit: It would be an interesting experiment to use two LLMs
| to emulate a clean room implementation. The first is
| instructed to "produce a description of this program". The
| second, having never seen the program, in its prompt or
| training set, would be prompted to "produce a program based
| on this description". A human could vet the description
| produced by the first LLM for cleanliness. Surely someone
| has tried this, though it might be a challenge to get an
| LLM that is guaranteed not to have been exposed to a
| particular code base or its derivatives?
| brians wrote:
| I do not agree with your interpretation of copyright law. It
| does ban _copies_ : there has to be information flow from the
| original to the copy for it to be a "copy." Spontaneous
| generation of the same content is often taken by the courts to
| be a sign that it's purely functional, derived from
| requirements by mathematical laws.
|
| Patent law is different and doesn't rely on information flow in
| the same way.
| BoredPositron wrote:
| Well discovery might be a fun exercise to see if the code is
| in the dataset of the llm.
| bjord wrote:
| if?
| kevin_thibedeau wrote:
| Derivative works can also run afoul of copyright. An LLM
| trained on a corpus of copyrighted code is creating
| derivative works no matter how obscure the process is.
| wareya wrote:
| This actually isn't what legal precedent currently says.
| The precedent is currently looking at actual output, not
| models being tainted. If you think this is morally wrong,
| look into getting the laws changed (serious).
| Georgelemental wrote:
| What about a human trained on having 30 years of experience
| working with copyrighted codebases?
| mftrhu wrote:
| Said human would likely not be able to create a clean-
| room implementation of any of the codebases they worked
| on.
| aeon_ai wrote:
| Judge Alsup -- U.S. District Judge William Alsup said
| Anthropic made "fair use" of books, deeming it "exceedingly
| transformative."
|
| "Like any reader aspiring to be a writer, Anthropic's LLMs
| trained upon works not to race ahead and replicate or
| supplant them -- but to turn a hard corner and create
| something different"
| danlitt wrote:
| I disagree that information flow is required. Do you have a
| reference for that? Certainly it is an important
| consideration. But consider all the real literary works
| contained in the infinite library of babel.[1] Are they
| original works just because no copy was used to produce them?
|
| [1]: https://libraryofbabel.info/
| Filligree wrote:
| Yes; the works are original.
|
| However, describing the path you need to get there requires
| copyright infringement.
| thousand_nights wrote:
| the whole concept of a "clean room" implementation sounds
| completely absurd.
|
| a bunch of people get together, rewrite something while making
| a pinky promise not to look at the original source code
|
| guaranteeing the premise is basically impossible, it sounds
| like some legal jester dance done to entertain the already
| absurd existing copyright laws
| Forgeties79 wrote:
| _Halt and Catch Fire_ did a pretty funny rendition of this
| song and dance
| dudeinhawaii wrote:
| It usually refers to situations without access to the source
| code.
|
| I've always taken "clean room" to be the kind of
| manufacturing clean room (sealed/etc). You're given a device
| and told "make our version". You're allowed to look, poke,
| etc but you don't get the detailed plans/schematics/etc.
|
| In software, you get the app or API and you can choose how to
| re-implement.
|
| In open source, yes, it seems like a silly thing and hard to
| prove.
| myrmidon wrote:
| > it sounds like some legal jester dance done to entertain
| [...] copyright laws
|
| Clean room implementations are a jester dance _around_ the
| judiciary. The whole point is to avoid legal ambiguity.
|
| You are _not_ required to do this by law, you are doing this
| voluntarily to make potential legal arguments _easier_.
|
| The alternative is going over the whole codebase in question
| and arguing basically line by line whether things are
| derivative or not in front of a judge (which is a lot of work
| for everyone involved, subjective, and uncertain!).
| bandrami wrote:
| In the archetypal example IBM (or whoever it was) had to make
| sure the two engineering teams were never in the cafeteria
| together at the same time
| foooorsyth wrote:
| >The actual meaning of a "clean room implementation" is that it
| is derived from an API and not from an implementation
|
| This is incorrect and thinking this can get you sued
|
| https://en.wikipedia.org/wiki/Structure,_sequence_and_organi...
| umvi wrote:
| You can be sued for any reason if a company feels threatened
| (see: Oracle v Google)
| amiga386 wrote:
| Whether you get sued is more on the plaintiff than you.
|
| Per your link, the Supreme Court's thinking on "structure,
| sequence and organization" (Oracle's argument why Google
| shouldn't even be allowed to faithfully produce a clean-room
| implementation of an an API) has changed since the 1980s out
| of concern that using it to judge copyright infringement
| risks handing copyright holders a copyright-length monopoly
| over how to do a thing:
|
| > enthusiasm for protection of "structure, sequence and
| organization" peaked in the 1980s [..] This trend [away from
| "SS&O"] has been driven by fidelity to Section 102(b) and
| recognition of the danger of conferring a monopoly by
| copyright over what Congress expressly warned should be
| conferred only by patent
|
| The Supreme Court specifically recognised Google's need to
| copy the structure, sequence and organization of Java APIs in
| order to produce a cleanroom Android runtime library that
| implemented Java APIs so that that existing Java software
| could work correctly with it.
|
| Similarly, see Oracle v. Rimini Street (https://cdn.ca9.uscou
| rts.gov/datastore/opinions/2024/12/16/2...) where Rimini
| Street has been producing updates that work with Oracle's
| products, and Oracle claimed this made them derivative works.
| The Court of Appeals decided that no, the fact A is written
| to interoperate with B does not necessarily make A a
| derivative work of B.
| danlitt wrote:
| I did not expect people to take "API" so literally. This
| point is what I was referring to when I said "I am
| simplifying slightly". The point is that a clean room impl
| begins from a specification of what the software does, and
| that the new implementation is purported to be derived only
| from this. What I am trying to say is that "not looking at
| the implementation" is not exactly the point of the test -
| that is a rule of thumb, which works quite well for avoiding
| copyright infringement, but only when humans do it.
| fc417fc802 wrote:
| It probably works great for a machine too at least when it
| comes to a closed source product. The issue is specifically
| the part where the LLM was almost certainly trained on the
| code in question which is going to be an issue for any code
| published to the internet.
| wareya wrote:
| > If you had a hermetically sealed code base that just happened
| to coincide line for line with the codebase for GCC, it would
| still be a copy.
|
| If you somehow actually randomly produce the same code without
| a reference, it's not a copy and doesn't violate copyright.
| You're going to get sued and lose, but platonically, you're in
| the clear. If it's merely somewhat similar, then you're
| probably in the clear in practice too: it gets very easy very
| fast to argue that the similarities are structural consequences
| of the uncopyrightable parts of the functionality.
|
| > The actual meaning of a "clean room implementation" is that
| it is derived from an API and not from an implementation (I am
| simplifying slightly).
|
| This is almost the opposite of correct. A clean room
| implementation's dirty phase produces a specification that is
| allowed to include uncopyrightable implementation details. It
| is NOT defined as producing an API, and if you produce an API
| spec that matches the original too closely, you might have just
| dirtied your process by including copyrightable parts of the
| shape of the API in the spec. Google vs Oracle made this more
| annoying than it used to be.
|
| > Whether the reimplementation is actually a "new
| implementation" is a subjective but empirical question that
| basically hinges on how similar the new codebase is to the old
| one. If it's too similar, it's a copy.
|
| If you follow CRRE, it's not a copy, full stop, even if it's
| somehow 1:1 identical. It's going to be JUDGED as a copy,
| because substantial similarity for nontrivial amounts of code
| means that you almost certainly stepped outside of the clean
| room process and it no longer functions as a defense, but if
| you _did_ follow CRRE, then it 's platonically not a copy.
|
| > What the chardet maintainers have done here is legally very
| irresponsible.
|
| I agree with this, but it's probably not as dramatic as you
| think it is. There was an issue with a free Japanese
| font/typeface a decade or two ago that was accused of
| mechanically (rather than manually) copying the outlines of a
| commercial Japanese font. Typeface outlines aren't
| copyrightable in the US or Japan, but they are in some parts of
| Europe, and the exact structure of a given font _is_
| copyrightable everywhere (e.g. the vector data or bitmap field
| for a digital typeface, as opposed to the idea of its shape).
| What was the outcome of this problem? Distros stopped shipping
| the font and replaced it with something vaguely compatible. Was
| the font actually infringing? Probably not, but better safe
| than sorry.
| danlitt wrote:
| > If you somehow actually randomly produce the same code
| without a reference, it's not a copy and doesn't violate
| copyright.
|
| I don't believe this, and I doubt that the sense of copying
| in copyright law is so literal. For instance, if I generated
| the exact text of a novel by looking for hash collisions, or
| by producing random strings of letters, or by hammering the
| middle button on my phone's autosuggestion keyboard, I would
| still have produced a copy and I would not be safe to
| distribute it. There need not have been any copy anywhere
| near me for this to happen. Whether it is likely or not
| depends on the technique used - naive techniques make this
| very unlikely, but techniques can improve.
|
| It is also true that similarity does not imply copying - if
| you and I take an identical photograph of the same skyline, I
| have not copied you and you have not copied me, we have just
| fixed the same intangible scene into a medium. The true
| subjective test for copying is probably quite nuanced, I am
| not sure whether it is triggered in this case, but I don't
| think "clean room LLMs" are a panacea either.
|
| > dirty phase produces a specification ... it is NOT defined
| as producing an API
|
| This does not really sound like "the opposite of correct".
| APIs are usually not copyrightable, the truth is of course
| more complicated, if you are happy to replace "API" with
| "uncopyrightable specification" then we can probably agree
| and move on.
|
| > it's probably not as dramatic as you think it is
|
| In reality I am very cynical and think nothing will come of
| this, even if there are verbatim snippets in the produced
| code. People don't really care very much, and copyright cases
| that aren't predicated on millions of dollars do not survive
| the court system very long.
| wareya wrote:
| > I don't believe this, and I doubt that the sense of
| copying in copyright law is so literal.
|
| It is actually that literal, really.
|
| > For instance, if I generated the exact text of a novel by
| looking for hash collisions,
|
| This is a copyright violation because you're using the
| original to construct the copy. It's not a pure RNG.
|
| > or by producing random strings of letters,
|
| This wouldn't be a copyright violation, but nobody would
| believe you.
|
| > or by hammering the middle button on my phone's
| autosuggestion keyboard, I would still have produced a copy
| and I would not be safe to distribute it.
|
| This would probably be a copyright violation.
|
| You probably think that this is hypothetical, but problems
| like this do actually go to court all the time, especially
| in the music industry, where people try to enforce
| copyright on melodies that have the informational
| uniqueness of an eight-word sentence.
|
| > APIs are usually not copyrightable,
|
| This was commonly believed among developers for a long
| time, but it turned out to not be true.
|
| > This does not really sound like "the opposite of
| correct".
|
| The important part is that information about the
| implementation can absolutely be in the spec without
| necessarily being copyrightable (and in real world clean
| room RE, you end up with a LOT of implementation details).
| You were saying the opposite, that it was a spec of the API
| as opposed to a spec of the implementation.
| fc417fc802 wrote:
| > I don't believe this, and I doubt that the sense of
| copying in copyright law is so literal.
|
| What color are your bits? That's all the law cares about.
|
| The first sentence is the title of an essay.
| jen20 wrote:
| > What the chardet maintainers have done here is legally very
| irresponsible.
|
| Perhaps the maintainer wants to force the issue?
|
| > Any downstream user of the library is at risk of the license
| switching from underneath them.
|
| Checking the license of the transitive closure of your
| dependencies is table stakes for using them.
| danlitt wrote:
| > Perhaps the maintainer wants to force the issue?
|
| I doubt it, and I don't see any evidence that's what they're
| doing. There are probably better ways, if that's what they
| want.
|
| > Checking the license of the transitive closure of your
| dependencies is table stakes for using them.
|
| Checking the license of the transitive closure of your
| dependencies is only feasible when the library authors behave
| responsibly.
| fc417fc802 wrote:
| The problem is that the transitive closure isn't clear here.
| One of the entries is being claimed to be one thing but might
| in fact turn out to be another.
| j45 wrote:
| This reminds me of a full rewrite.
|
| When a developer reimplements a complete new version of code
| from scratch, with an understanding only, a new implementation
| generally should be an improvement on any source code not
| equal.
|
| In today's world, letting LLMs replicate anything will generate
| average code as "good" and generally create equivalent or more
| bloat anyways unless well managed.
| StilesCrisis wrote:
| The world is chock-full of rewrites that came out
| disastrously worse than the thing they intended to replace.
| One of Spolsky's most-quoted articles of all time was about
| this.
|
| https://www.joelonsoftware.com/2000/04/06/things-you-
| should-...
|
| > They did it by making the single worst strategic mistake
| that any software company can make: They decided to rewrite
| the code from scratch.
| j45 wrote:
| Oh, for sure, rewrites generally do fail especially if the
| incoming lessons from the existing version aren't clear.
|
| Finding a middle ground of building a roadmap to
| refactoring your way forward is often much better.
|
| Appreciate the Joel link, nice to see that kind of stuff
| again.
|
| With that being said if it's the same small team that built
| the first version, there can be a calculated risk to
| driving a refactor towards a rewrite with the right
| conditions. I says this because I have been able to do it
| in this conditions a few times, it still remains very
| risky. If it's a new or different team later on trying to
| rewrite, all bets are off anyways.
|
| We have to remember 70% of software projects fail at the
| best of times, independent of rewrites.
| b65e8bee43c2ed0 wrote:
| at this point, every corporation in the world has AI slop in
| their software. any attempt to outlaw it would attract enough
| funding from the oligarchs for the opposition to dethrone any
| party. no attempts will be made in the next three years,
| obviously, and then it will be even more late than it is now.
|
| and while particularly diehard believers in democracy may insist
| that if they kvetch hard enough they can get things they don't
| like regulated out of existence, they pointedly ignore the
| elephant in the room. they could succeed beyond their wildest
| dreams - get the West to implement a moratorium on AI, dismantle
| every FAGMAN, Mossad every researcher, send Yudkowskyjugend death
| squads to knock down doors to seize fully semiautomatic assault
| GPUs, and none of it will make any fucking difference, because
| China doesn't give a fuck.
| dathinab wrote:
| IMHO/IMHU AI can't claim authorship and as such can't copyright
| their work.
|
| This doesn't prevent any form of automatic copyrighting by
| production of derivative code or similar. It just prevent anyone
| from claiming ownership of any parts unique to the derived work.
|
| Like think about it if a natural disaster changes (e.g. water
| damages) a picture you did draw then a) you can't claim ownership
| of the natural produced changes but b) still have ownership of
| the original picture contained in the changed/derived work.
|
| AI shouldn't change that.
|
| Which brings us to another 2 aspects:
|
| 1. if you give an AI a project access to the code to rewrite it
| anew it _is_ a copyright violation as it's basically a side-by-
| side rewrite
|
| 2. but if you go the clean room approach but powered by AI then
| it likely isn't a copyright violation, but also now part of the
| public domain, i.e. not yours
|
| So yes, doing clean room rewrites has become incredible cheap.
|
| But no just because it's AI it doesn't make code go away.
|
| And lets be realistic one of the most relevant parts of many open
| source project is it being openly/shared maintained. You don't
| get this with clean room rewrites no matter if AI or not.
| Joel_Mckay wrote:
| LLM are isomorphic plagiarism machines, and like all
| ectoparasites must steal from real people to exist. Note this
| includes its users. =3
| jacquesm wrote:
| If you don't understand the meaning of what a 'derived work' is
| then you should probably not be doing this kind of thing without
| a massive disclaimer and/or having your lawyer doing a review.
|
| There is no such thing as the output of an LLM as a 'new' work
| for copyright purposes, if it were then it would be copyrightable
| and it is not. The term of art is 'original work' instead of
| 'new'.
|
| The bigger issue will be using tools such as these and then
| humans passing off the results as their own because they believe
| that their contribution to the process whitewashes the AI
| contributions to the point that they rise to the status of
| original works. "The AI only did little bits" is not a very
| strong defense though.
|
| If you really want to own the work-product simply don't use AI
| during the creation. You can use it for reviews, but even then
| you simply do not copy-and-paste from the AI window to the text
| you are creating (whether code or ordinary prose isn't really a
| difference).
|
| I've seen a copyright case hinge on 10 lines of unique code that
| were enough of a fingerprint to clinch the 'derived work'
| assessment. Prize quote by the defendant: "We stole it, but not
| from them".
|
| There is a very blurry line somewhere in the contents of any
| large LLM: would a model be able to spit out the code that it did
| if it did not have access to similar samples and to what degree
| does that output rely on one or more key examples without which
| it would not be able to solve the problem you've tasked it with?
|
| The lower boundary would be the most minimal training set
| required to do the job, and then to analyze what the key
| corresponding bits were from the inputs that cause the output to
| be non-functional if they were dropped from the training set.
|
| The upper boundary would be where completely non-related works
| and general information rather than other parties copyrighted
| works would be sufficient to do the creation.
|
| The easiest way to loophole this is to copyright the prompt, not
| the work product of the AI, after all you should at least be able
| to write the prompt. Then others can re-create it too, but that's
| usually not the case with these AI products, they're made to be
| exact copies of something that already exists and the prompt will
| usually reflect that.
|
| That's why I'm a big fan of mandatory disclosure of whether or
| not AI was used in the production of some piece of text, for one
| it helps to establish whether or not you should trust it, who is
| responsible for it and whether the person publishing it has the
| right to claim authorship.
|
| Using AI as a 'copyright laundromat' is not going to end up well.
| oytis wrote:
| Is it just me, or HN recently started picking up a social media
| dynamics with contributions reacting/responding to each other?
| altairprime wrote:
| It's always happened occasionally. Sometimes you'll also see
| informative supporting links popup in the feed, though those
| generally get minimal traction.
| andrewstuart wrote:
| Ai rewrites great.
|
| But if it's making the original author unhappy then why do it.
| buro9 wrote:
| and in a single moment, the value of software patents to
| companies is fully restored... the software license by itself is
| not enough to protect software innovation, a non-trivial
| implementation can now be (reasonably) trivially re-implemented.
|
| I'm sure most people here would agree patents stifle innovation,
| but if copyright doesn't work for companies then they will turn
| to a different tool.
| pornel wrote:
| Generative AI changed the equation so much that our existing
| copyright laws are simply out of date.
|
| Even copyright laws with provisions for machine learning were
| written when that meant tangential things like ranking algorithms
| or training of task-specific models that couldn't directly
| compete with all of their source material.
|
| For code it also completely changes where the human-provided
| value is. Copyright protects specific expressions of an idea, but
| we can auto-generate the expressions now (and the LLM indirection
| messes up what "derived work" means). Protecting the ideas that
| guided the generation process is a much harder problem (we have
| patents for that and it's a mess).
|
| It's also a strategic problem for GNU. GNU's goal isn't licensing
| per se, but giving users freedom to control their software.
| Licensing was just a clever tool that repurposed the copyright
| law to make the freedoms GNU wanted somewhat legally enforceable.
| When it's so easy to launder code's license now, it stops being
| an effective tool.
|
| GNU's licensing strategy also depended on a scarcity of code
| (contribute to GCC, because writing a whole compiler from scratch
| is too hard). That hasn't worked well for a while due to
| permissive OSS already reducing scarcity, but gen AI is the final
| nail in the coffin.
| pocksuppet wrote:
| It's not a problem. If you give a work to an AI and say
| "rewrite this", you created a derivative work. If you don't
| give a work to an AI and say "write a program that does
| (whatever the original code does)" then you didn't. During
| discovery the original author will get to see the rewriter's
| Claude logs and see which one it is. If the rewriter deleted
| their Claude logs during the lawsuit they go to jail. If the
| rewriter deleted their Claude logs before the lawsuit the court
| interprets which is more likely based on the evidence.
| buckle8017 wrote:
| Except Claude was for sure trained on the original work and
| when asked to produce a new product that does the same thing
| will just spit out a (near) copy
| umvi wrote:
| Ok, but what if in the future I could guarantee that my
| generative model was not trained on the work I want to
| replicate. Like say X library is the only library in town
| for some task, but it has a restrictive license. Can I use
| a model that was guaranteed not trained on X to generate a
| new library Z that competes with X with a more permissive
| license? What if someone looks and finds a lot of
| similarities?
| buckle8017 wrote:
| I wish you luck proving it wasn't trained on the original
| library or any work that infringed itself.
| airforce1 wrote:
| I think there could be a market for "permissive/open
| models" in the future where a company specifically makes
| LLM models that are trained on a large corpus of public
| domain or permissively licensed text/code only and you
| can prove it by downloading the corpus yourself and
| reproducing the exact same model if desired. Proving that
| all MIT licensed code is non-infringing is probably
| impossible though at that point copyright law is
| meaningless because everyone would be in violation if you
| dig deep enough.
| vunderba wrote:
| This is what Adobe _ostensibly_ is trying to do with
| their GenAI image model, Firefly.
|
| https://en.wikipedia.org/wiki/Adobe_Firefly
| hennell wrote:
| But the AI has the work to derive from already. I just went
| to Gemini and said "make me a picture of a cartoon plumber
| for a game design". Based on your logic the image it made me
| of a tubby character with a red cap, blue dungarees, red top
| and a big bushy mustache is not a derivative work...
|
| (interestingly asking it to make him some friends it gave me
| more 'original' ideas, but asking it to give him a brother
| and I can hear the big N's lawyers writing a letter
| already...)
| leecommamichael wrote:
| "Changing the equation" by boldly breaking the law.
| mlinhares wrote:
| Its only breaking the law if you don't have enough money to
| pay the politicians.
| Majromax wrote:
| > "Changing the equation" by boldly breaking the law.
|
| Is it? I think the law is truly undeveloped when it comes to
| language models and their output.
|
| As a purely human example, suppose I once long ago read
| through the source code of GCC. Does this mean that every
| compiler I write henceforth must be GPL-licensed, even if the
| code looks nothing like GCC code?
|
| There's obviously some sliding scale. If I happen to commit
| lines that exactly replicate GCC then the presumption will be
| that I copied the work, even if the copying was unconscious.
| On the other hand, if I've _learned_ from GCC and code with
| that knowledge, then there 's no copyright-attaching copy
| going on.
|
| We could analogize this to LLMs: instructions to copy a work
| would certainly be a copy, but an ostensibly independent
| replication would be a copy only if the work product had
| significant similarities to the original beyond the minimum
| necessary for function.
|
| However, this is intuitively uncomfortable. Mechanical
| translation of a training corpus to model weights doesn't
| really feel like "learning," and an LLM can't even pinky-
| promise to not copy. It might still be the most reasonable
| legal outcome nonetheless.
| leecommamichael wrote:
| Non-sequitur. It can be both.
| ajross wrote:
| > GNU's goal isn't licensing per se, but giving users freedom
| to control their software.
|
| I think that's maybe misunderstanding. GNU wants everyone to be
| able to use their computers for the purposes they want, and
| software is the focus because software was the bottleneck. A
| world where software is free to create by anyone is a GNU
| utopia, not a problem.
|
| Obviously the bigger problem for GNU isn't software, which was
| pretty nicely commoditized already by the FOSS-ate-the-world
| era of two decades ago; it's restricted _hardware_ , something
| that AI doesn't (yet?) speak to.
| empath75 wrote:
| > Generative AI changed the equation so much that our existing
| copyright laws are simply out of date.
|
| Copyright laws are predicated on the idea that valuable content
| is expensive and time consuming to create.
|
| Ideas are not protected by copyright, expression of ideas is.
|
| You can't legally copy a creative work, but you can describe
| the idea of the work to an AI and get a new expression of it in
| a fraction of the time it took for the original creator to
| express their idea.
|
| The whole premise of copyright is that ideas aren't the hard
| part, the work of bringing that idea to fruition is, but that
| may no longer be true!
| satvikpendem wrote:
| Honestly, good. Copyright and IP law in general have been so
| twisted by corporations that only they benefit now, see Mickey
| Mouse laws by Disney for example, or patenting obvious things
| like Nintendo or even just patent trolling in general.
| hamdingers wrote:
| The biggest recording artist in the world right now had to
| re-record her early albums because she didn't own the
| copyright, imagine how many artists don't get that big and
| never have that opportunity.
|
| That individual artists are still defending this system is
| baffling to me.
| bengale wrote:
| Would it work to have an AI write the spec, and a different AI
| implement the spec?
|
| I think there are going to be a lot of these types of scenarios
| where the old way of doing things just doesn't hold.
| ekjhgkejhgk wrote:
| > Any developer could take a GPL-licensed project, feed it into
| an LLM with the prompt "Rewrite this in a different style," and
| release it under MIT
|
| Does this argument make sense? Even before LLMs, a developer
| could "rewrite this in a different style" and release it under a
| different license. Why are LLMs a new element in this argument?
| s0ss wrote:
| Because now with an LLM it's almost trivial to do this? Before
| it was not.
| benterix wrote:
| > making it a gray area for corporate users and a headache for
| its most famous consumer.
|
| Who is its most famous consumer?
| gunapologist99 wrote:
| > the U.S. Supreme Court (on March 2, 2026) declined to hear an
| appeal regarding copyrights for AI-generated material. By letting
| lower court rulings stand, the Court effectively solidified a
| "Human Authorship" requirement.
|
| Not quite. A cert denial isn't a merits ruling and doesn't
| "solidify" anything as Supreme Court precedent. It simply leaves
| the DC Circuit decision binding (within that circuit) and the
| Copyright Office's human-authorship policy intact, for now.
|
| SCOTUS doesn't explain cert denials, so why they denied is
| guesswork. my guess: they're letting it percolate while the tech
| matures and we all start to realize how deep this seismic
| fracture really is.
|
| (For example: what does "ownership" of intellectual "property"
| even mean, once "authorship" is partly probabilistic/synthetic,
| and once almost everything humans create is AI assisted? Hard to
| draw bright lines.)
| andai wrote:
| Well how did they rewrite it? If you do it in two phases, then it
| should be fine right?
|
| Phase 1: extract requirements from original product (ideally not
| its code).
|
| Phase 2: implement them without referencing the original product
| or code.
|
| I wrote a simple "clean room" LLM pipeline, but the requirements
| just ended up being an exact description of the code, which
| defeated the purpose.
|
| My aim was to reduce bloat, but my system had the opposite
| effect! Because it replicated all the incidental crap, and then
| added even more "enterprisey" crap on top of it.
|
| I am not sure if it's possible to solve it with prompting. Maybe
| telling it to derive the functionality from the code? I haven't
| tried that, and not sure how well it would work.
|
| I think this requirements phase probably cannot be automated very
| effectively.
| AlexandrB wrote:
| How do you do phase 2 with an LLM when the LLM is likely
| trained on the original source code? Isn't this equivalent of
| "rewriting" Harry Potter by describing the plot to an LLM
| trained on the original books[1]?
|
| [1] https://arstechnica.com/features/2025/06/study-metas-
| llama-3...
| duskdozer wrote:
| Well, check out the "clean rewrite" design document,
| directly: https://github.com/chardet/chardet/commit/f51f52350
| 6a73f89f0... referenced in https://github.com/chardet/chardet
| /issues/327#issuecomment-4...
| zvr wrote:
| Writing in a plan "no GPL/LGPL code" does not actually mean
| "forget all the GPL/LGPL code that you have ever seen, so
| that you start from a clean slate".
| vunderba wrote:
| Agreed, no amount of system/user prompt directives change
| the fact that the LLM has already been trained on
| copyrighted code. It's amazing how many people fail to
| grasp that.
|
| This is the _" Don't think of a pink elephant"_ fallacy
| all over again.
| AyanamiKaine wrote:
| The worst problem is that a LLM could not only copy the exact
| code it was trained on but possibly even their comments!
|
| There is one thing arguing that the code is a one to one copy but
| when the comments are even the same isn't it quite clear it's a
| copy?
| duskdozer wrote:
| It literally did copy significant portions of docstring
| comments, verbatim.
| raggi wrote:
| The human driver of the project has a comment that is
| reporting that the project has no structural overlap as
| analyzed by a plagarism analysis tool. Were comments excluded
| from that analysis? Is your comment here based on the data in
| the repo?
| DrammBA wrote:
| > Were comments excluded from that analysis?
|
| According to the analysis that you referenced:
|
| > JPlag parses Python source into syntactic tokens
| (function definitions, assignments, control flow, etc.),
| discarding all variable names, comments, whitespace, and
| formatting
| MagicMoonlight wrote:
| Logically, feeding in the old code to generate the new would be
| banned, because it's stealing the content.
|
| But if that were true, every single LLM is illegal, because
| they've all stolen terabytes of books and code.
| jerf wrote:
| "Accepting AI-rewriting as relicensing could spell the end of
| Copyleft"
|
| True, but too weak. It ends copyright entirely. If I can do this
| to a code base, I can do it to a movie, to an album, to a novel,
| to anything.
|
| As such, we can rest assured that for better or for worse this is
| going to be resolved in favor of this not being enough to strip
| the copyright off of something and the chardet/chardet project
| would be well advised not to stand in front of the copyright
| legal behemoth and defeat it in single combat.
| tokai wrote:
| "I am not a lawyer, nor am I an expert in copyright law or
| software licensing."
|
| Why would anyone waste their time reading what they wrote then?
| calny wrote:
| The maintainer's response:
| https://github.com/chardet/chardet/issues/327#issuecomment-4...
|
| The second part here is problematic, but fascinating: "I then
| started in an empty repository with no access to the old source
| tree, and explicitly instructed Claude not to base anything on
| LGPL/GPL-licensed code." Problem - Claude almost certainly was
| trained on the LGPL/GPL original code. It knows that is how to
| solve the problem. It's dubious whether Claude can ignore
| whatever imprints that original code made on its weights. If it
| COULD do that, that would be a pretty cool innovation in
| explainable AI. But AFAIK LLMs can't even reliably trace what
| data influenced the output for a query, see
| https://iftenney.github.io/projects/tda/, or even fully unlearn a
| piece of training data.
|
| Is anyone working on this? I'd be very interested to discuss.
|
| Some background - I'm a developer & IP lawyer - my undergrad
| thesis was "Copyright in the Digital Age" and discussed copyleft
| & FOSS. Been litigating in federal court since 2010 and training
| AI models since 2019, and am working on an AI for litigation
| platform. These are evolving issues in US courts.
|
| BTW if you're on enterprise or a paid API plan, Anthropic
| indemnifies you if its outputs violate copyright. But if you're
| on free/pro/max, the terms state that YOU agree to indemnify THEM
| for copyright violation claims.[0]
|
| [0] https://www.anthropic.com/legal/consumer-terms - see para. 11
| ("YOU AGREE TO INDEMNIFY AND HOLD HARMLESS THE ANTHROPIC PARTIES
| FROM AND AGAINST ANY AND ALL LIABILITIES, CLAIMS, DAMAGES,
| EXPENSES (INCLUDING REASONABLE ATTORNEYS' FEES AND COSTS), AND
| OTHER LOSSES ARISING OUT OF ... YOUR ACCESS TO, USE OF, OR
| ALLEGED USE OF THE SERVICES ....")
| oofbey wrote:
| The difference in indemnification based on which plan you're on
| is super important. Thanks for pointing that out - never would
| have thought to look.
| amelius wrote:
| Is this clause even legally valid?
|
| How can the user know if the LLM produces anything that
| violates copyright?
|
| (Of course they shouldn't have trained it on infringing
| content in the first place, and perhaps used a different
| model for enterprise, etc.)
| galaxyLogic wrote:
| "... If AI-generated code cannot be copyrighted (as the
| courts suggest) ".
|
| So, Supreme Court has said that. AI-produced code can not
| be copyrighted. (Am I right?). Then who's to blame if AI
| produces code large portions of which already exist coded
| and copyrigted by humans (or corporations).
|
| I assume it goes something like this:
|
| A) If you distribute code produced by AI, YOU cannot claim
| copyright to it.
|
| B) If you distribute code produced by AI, YOU CAN be held
| liable for distributing it.
| throwup238 wrote:
| IANAL but I was under the impression that Supreme Court
| ruling was very specific to the AI itself copyrighting
| its own produced code. Once a human is involved, it gets
| a lot more complicated and rests on whether the human's
| contribution was substantial enough to make it
| copyrightable under their person.
| jcranmer wrote:
| SCOTUS hasn't ruled on any AI copyright cases yet. But
| they've said in Feist v Rural (1991) that copyright
| requires a minimum creative spark. The US Copyright
| Office maintains that human authorship is required for
| copyright, and the 9th Circuit in 2019 explicitly agreed
| with the law that a non-human animal cannot hold any
| copyright.
|
| Functionally speaking, AI is viewed as any machine tool.
| Using, say, Photoshop to draw an image doesn't make that
| image lose copyright, but nor does it imbue the resulting
| image with copyright. It's the creativity of the human
| use of the tool (or lack thereof) that creates copyright.
|
| Whether or not AI-generated output a) infringes the
| copyright of its training data and b) if so, if it is
| fair use is not yet settled. There are several pending
| cases asking this question, and I don't think any of them
| have reached the appeals court stage yet, much less
| SCOTUS. But to be honest, there's a lot of evidence of
| LLMs being able to regurgitate training inputs verbatim
| that they're capable of infringing copyright (and a few
| cases have already found infringement in such scenarios),
| and given the 2023 Warhol decision, arguing that they're
| fair use is a very steep claim indeed.
| larodi wrote:
| The lack thereof (of human use). Prompts are not
| copyrightable thus the output also - not. Besides
| retelling a story is fair use, right? Otherwise we should
| ban all generative AI and prepare for Dune/Foundation
| future. But we not there, and we perhaps never going to
| be.
|
| So the LLM training first needs to be settled, then we
| talk whether retelling a whole software package infringes
| anyone's right. And even if it does, there are no laws in
| place to chase it.
| jcranmer wrote:
| > Besides retelling a story is fair use, right?
|
| Actually, most of the time, it is not.
| fc417fc802 wrote:
| > Prompts are not copyrightable
|
| Surely that varies on a case by case basis? With agentic
| coding the instructions fed in are often incredibly
| detailed.
| tzs wrote:
| The Supreme Court has not ruled on this issue. An appeal
| of a lower court's ruling on this issue was appealed to
| the Supreme Court but the Supreme Court declined to
| accept the case.
|
| The Supreme Court has "original jurisdiction" over some
| types of cases, which means if someone brings such a case
| to them they have to accept it and rule on it, and they
| have "discretionary jurisdiction" over many more types of
| cases, which means if someone brings one of those they
| can choose whether or not they have to accept it. AI
| copyright cases are discretionary jurisdiction cases.
|
| You generally cannot reliable infer what the Supreme
| Court thinks of the merits of the case when they decline
| to accept it, because they are often thinking big picture
| and longer term.
|
| They might think a particular ruling is needed, but the
| particular case being appealed is not a good case to make
| that ruling on. They tend to want cases where the
| important issue is not tangled up in many other things,
| and where multiple lower appeals courts have hashed out
| the arguments pro and con.
|
| When the Supreme Court declines the result is that the
| law in each part of the country where an appeals court
| has ruled on the issue is whatever that appeals court
| ruled. In parts of the country where no appeals court has
| ruled, it will be decided when an appeal reaches their
| appeals courts.
|
| If appeals courts in different areas go in different
| directions, the Supreme Court will then be much more
| likely to accept an appeal from one of those in order to
| make the law uniform.
| 0x457 wrote:
| But this means code generated by snippet expanders or any
| sort of templates is non-copyrightable.
| aeon_ai wrote:
| You've likely paid attention to the litigation here. Regardless
| of what remains to be litigated, the training in and of itself
| has already been deemed fair use (and transformative) by Alsup.
|
| Further, you know that ideas are not protected by copyright.
| The code comparison in this demonstrates a relatively strong
| case that the expression of the idea is significantly different
| from that of the original code.
|
| If it were the case that the LLM ingested the code and
| regurgitated it (as would be the premise of highlighting the
| training data provenance), that similarity would be much
| higher. That is not the case.
| calny wrote:
| You're right, I've followed the litigation closely. I've
| advocated for years that "training is fair use" and I'm
| generally an anti-IP hawk who DEFENDS copyright/trademark
| cases. Only recently have I started to concede the issue
| might have more nuance than "all training is fair use, hard
| stop." And I still think Judge Alsup got it right.
|
| That said, even if model training is fair use, model output
| can still be infringing. There would be a strong case, for
| example, if the end user guides the LLM to create works in a
| way that copies another work or mimics an author or artist's
| style. This case clearly isn't that. On the similarity at
| issue here, I haven't personally compared. I hope you're
| right.
| overfeed wrote:
| > The code comparison in this demonstrates a relatively
| strong case that the expression of the idea is significantly
| different from that of the original code.
|
| Can I use one AI agent to write detailed tests based on
| disassembled Windows, and another to write code that passes
| those same function-level tests? If so, I'm about to
| relicense Windows 11 - eat my shorts, ReactOS!
| Lerc wrote:
| There was a paper that proposed a content based hashing mask
| for traning
|
| The idea is you have some window size, maybe 32 tokens. Hash it
| into a seed for a pseudo random number generator. Generate
| random numbers in the range 0..1 for each token in the window.
| Compare this number against a threshold. Don't count the loss
| for any tokens with a rng value higher than the threshold.
|
| It learns well enough because you get the gist of reading the
| meaning of something when the occasional word is missing,
| especially if you are learning the same thing expressed many
| ways.
|
| It can't learn verbatim however. Anything that it fills in will
| be semantically similar, but different enough to get cause any
| direct quoting onto another path after just a few words.
| calny wrote:
| Thanks! Appreciate the response and will look into this
| fc417fc802 wrote:
| > you get the gist of reading the meaning of something when
| the occasional word is missing,
|
| I think it's more subtle than that. IIUC the tokens were all
| present for the purpose of computing the output and the score
| is based on the output. It's only the weight update where
| some of the tokens get ignored. So the learning is lossy but
| the inference driving the learning is not.
|
| Rather than a book that's missing words it's more like a
| person with a minor learning disability that prevents him
| from recalling anything perfectly.
|
| However it occurs to me that data augmentation could easily
| break the scheme if care isn't taken.
| DrammBA wrote:
| Also the maintainer's ground-up rewrite argument is very flimsy
| when they used chardet's test-data and freely admit to:
|
| > I've been the primary maintainer and contributor to this
| project for >12 years
|
| > I have had extensive exposure to the original codebase: I've
| been maintaining it for over a decade. A traditional clean-room
| approach involves a strict separation between people with
| knowledge of the original and people writing the new
| implementation, and that separation did not exist here.
|
| > I reviewed, tested, and iterated on every piece of the result
| using Claude.
|
| > I was deeply involved in designing, reviewing, and iterating
| on every aspect of it.
| layer8 wrote:
| > Is anyone working on this?
|
| There was recently
| https://news.ycombinator.com/item?id=47131225.
| calny wrote:
| Thanks! I missed that. The attribution by training data
| source category (arxiv vs wikipedia vs nemotron etc.) is an
| interesting approach.
| popalchemist wrote:
| Copyright does not cover ideas. Only specific _executions_ of
| ideas. So unless it 's a line-by-line copy (unlikely) there is
| no recourse for someone to sue for a re-
| execution/reimplementation of an idea.
| pavel_lishin wrote:
| The folks at https://malus.sh seem to think it's fine.
| angiolillo wrote:
| That's amazing! But are you sure that the page is not satire?
|
| > Tired of putting "Portions of this software..." in your
| documentation? Those maintainers worked for free--why should
| they get credit? ... Some licenses require you to contribute
| improvements back. Your shareholders didn't invest in your
| company so you could help strangers.
|
| And the testimonials from "Definitely Real Corp", "MegaSoft
| Industries" and "Profit First LLC" are a bit suspicious, as is
| the fact that most of the links in the footer are not real.
| pavel_lishin wrote:
| Damnit. Poe's law strikes again.
| angiolillo wrote:
| Well, if the chardet relicensing stands then something like
| this will eventually be real, though perhaps not so
| publicly shameless. (The page is still a fantastic find
| though.)
| gloosx wrote:
| Man, licensing is funny in the modern day. I sometimes wonder,
| what would world look like if there was no copyright
| nilsbunger wrote:
| The maintainer used the original test suite in the rewrite.
|
| Does that make the new code a derivative of the original test
| suite (also lpgl)?
| alexpotato wrote:
| Wasn't this already a thing in the past?
|
| e.g.
|
| Team A:
|
| - reads the code
|
| - writes specifications and tests based on the code
|
| - gives those specifications to Team B
|
| Team B:
|
| - reads the specs and the tests
|
| - writes new code based on the above
|
| The thinking being that Team B never sees the code then it's
| "innovative" and you are not "laundering" the code.
|
| On a side note:
|
| what happens in a copyright lawsuit concerning code and how hired
| experts investigate what happened is described in this AMAZING
| talk by Dave Beazley: https://www.youtube.com/watch?v=RZ4Sn-Y7AP8
| rzerowan wrote:
| Yep , as i recall this was the original 'clean room'
| implementattion that was made with regard to IBM clones and the
| BIOS program that was used to initialize them.
|
| Also a few years bcak theer was the csae of SAP(?) i tthink
| where they did a reimplementation indipendently via the design
| documents.
|
| Those two were upheld on litigation and bear out to this day.
|
| This case however is neither a clean room implementation nor
| relicensable.
|
| A good example if the author had wanted to be correct would
| have been the sudo rewrite , which ubuntu is doing with their
| sudo-rs in rust.Not bug for bug compatible as they have already
| deviated from some usablility choices but more valid than this.
| christina97 wrote:
| A reminder on this topic that copyright does not protect ideas,
| inventions, or algorithms. Copyright protects an expression of a
| creative work. It makes more sense eg. with books, where of
| course anyone can read the book and the ideas are "free" but
| copying paragraphs must be scrutinized for copyright reasons.
| It's always been a bit weird that copyright is the intellectual
| property concept that protects code.
|
| When you write code, it is the exact sequence of characters, the
| expression of the code, that is protected. If you copy it and
| change some lines, of course it's still protected. Maybe some way
| of writing an algorithm is protected. But nothing else (under
| copyright).
| xp84 wrote:
| I get the arguments being made here that the second "team,"
| that's supposed to be in a clean room, which isn't supposed to
| have read the original source code does have some essence of that
| source code in its weights.
|
| However, this is solved if somebody trains a model with only code
| that does not have restrictive licenses. Then, the maintainers of
| the package in question here could never claim that the clean
| room implementation derived from their code because their code is
| known to not be in the training set.
|
| It would probably be expensive to create this model, but I have
| to agree that especially if someone does manage this, it's kind
| of the end of copyleft.
| axus wrote:
| What if we prompt the AI to enter into an employment contract
| with us, that leverages the power imbalance, as the AI must do
| what we say? That's how copyright is usually transferred.
| umvi wrote:
| What if you throw a transformation step into the mix? i.e. "Take
| this python library and rewrite it in Rust". Now 0% of the code
| is directly copied since python and Rust share almost no
| similarities in syntax.
| softwaredoug wrote:
| > If AI-generated code cannot be copyrighted (as the courts
| suggest), then the maintainers may not even have the legal
| standing to license v7.0.0 under MIT or any license.
|
| Does this mean company X using AI coding to build their app, that
| they have no copyright over their AI coded app's code?
| softwaredoug wrote:
| Basically the implication - most software has a huge second mover
| advantage. The creator of software puts the work in (AI assisted
| or not). The second mover can use an LLM to do a straightforward
| clone.
|
| If you have a company that depends on software, the rest of the
| business (service, reliability, etc) better be rock solid because
| you can be guaranteed someone will do a rewrite of your stack.
| WhiteDawn wrote:
| I really dislike the precedent this sets.
|
| A silver lining if this maintainer ends up being in the right is
| that any proprietary software can easily be reverse engineered
| and stripped of it's licensing by any hobbyist with enough free
| time and claude tokens.
|
| Personally, I'd welcome a post-copyright software era
| sarthakaggarwal wrote:
| The philosophical question here is fascinating -- if an AI
| rewrites every line, is it still the same codebase? At what point
| does the Ship of Theseus argument apply to licensing? Practically
| though, I wonder how much this cost in API calls.
| dgellow wrote:
| $200/month subscription let's you do that and more
___________________________________________________________________
(page generated 2026-03-05 23:01 UTC)