Posts by lucy@starlightnet.work
(DIR) Post #B51b6yqBbafDIZzy5I by lucy@starlightnet.work
0 likes, 0 repeats
@alexia WHAT??? ..is the pizza gold plated?Is it BLE capable? For that price, it better be BLE capable
(DIR) Post #B5DgLKTemxZq7pdoZc by lucy@starlightnet.work
1 likes, 1 repeats
@lumi i've said my piece as well.krita-artists.org/t/policy-on-llm-code/178248/36?u=thelucyIf you care about AI use in the Krita codebase, make your voice heard before the thread dies down
(DIR) Post #B5DgLLRDDXDR6XVNOy by lucy@starlightnet.work
0 likes, 0 repeats
"why are you so anti-ai"because our future generations will not forgive us for our apathy. thats why. thats why i'm doing any of this. that's why i joined tsn and started making hardware.it's worth at least trying.
(DIR) Post #B5FjFT1km9Kr5FPDfM by lucy@starlightnet.work
0 likes, 0 repeats
Nothin to see here... #fckai
(DIR) Post #B5FjFTEW0hXlipNPYe by lucy@starlightnet.work
1 likes, 1 repeats
"cyanide" is an LLM defense tool I'm working on. It's meant to poison-pill your data so an LLM can't work with it at inference-time.Nothing ready for prod yet, i'm still experimenting with different things. From: @lucyRE: shrimp.starlightnet.work/notes/al0n7ifaqy7076jw
(DIR) Post #B5K4ywZsd6jLQwjZui by lucy@starlightnet.work
0 likes, 0 repeats
@libreleah how do you cope without line numbers ðŸ˜
(DIR) Post #B5K5FiukgPNLyqUvpo by lucy@starlightnet.work
0 likes, 0 repeats
@libreleah thats actually a pretty good reason nglbest of luck with librewolf \^\^
(DIR) Post #B5K5eWoCpPzZfvGQ9w by lucy@starlightnet.work
0 likes, 0 repeats
@libreleah oh really?Damn, I use it all the time. I use relative line numbers and jump around the codebase with e.g. "8j" quite frequently, especially at work.
(DIR) Post #B5PwegimEzDRmMUAHA by lucy@starlightnet.work
1 likes, 2 repeats
We gotta fight back. Contacting other open source projects and asking them to (hopefully) adopt a no-AI policy is a good first step, but we need to go further.We need to make it harder for users to use AI with our content, and we need to make it harder for these companies to steal our shit.I'm working on a tool, called cyanide, in order to poison LLMs at inference-level so they're absolutely useless when told to summarise content from a given website.I encourage you to do the same. figure out ways to poison and otherwise break LLMs when they deal with your content. At inference level, at training level, it doesn't matter.The more people do this, the more diverse tooling we build to stop these AI bro fucks from stealing our shit, the harder it is for the AI companies to clean up our shit, and the more obvious the failures and shortcomings of LLMs as search engines become for the average user.Here are a couple resources I found helpful when researching LLM security topics:- promptfoo.dev, a database of LLM vulnerabilities with links to research papers, which models are affected, etc.- [google scholar, since these things are an active field of research](scholar.google.com)- [openrouter.ai, cheap easy testing, especially when tested against multiple models since it's one simple API](openrouter.ai)feel free to ping me for resources to add, and other tooling to break LLMs with the data they scrape, parse or train on itself (so while iocaine is cool and helpful, it doesn't really poison inputdata itself, it feeds different data depending on who accesses the page)
(DIR) Post #B5aT9oOKue6oevPOM4 by lucy@starlightnet.work
1 likes, 0 repeats
I'm so tired of all the AI inevitability bullshit I hear from a lot of people.Somehow it usually turns out to be users who are gladly using it.
(DIR) Post #B5aUPHAaIXy2D8ujFA by lucy@starlightnet.work
1 likes, 0 repeats
"Like it or not, AI is here to stay".Unless the frontier AI companies are going under because it is yet to turn a profit.Unless we prevent AI from accessing our web content for scraping or inference.Unless any sort of regulation/law, be it existing or hypothetical ones in the future, regulates what's allowed in the training data.Unless the workers of the global south who sort out and label the training data unionize, go on strike, or find otherwise better paid jobs at some point in the future.Unless they hit a limit of how much power they can pull at once in a datacenter.Unless users actually learn and understand how the tech behind it actually works.Unless we refuse to play a part in their bs and just refuse to use it, collectively.Unless [...]Unless [...]Unless [...]
(DIR) Post #B5iuqH5qqGR1rjujjs by lucy@starlightnet.work
1 likes, 0 repeats
Talk is cheap, send code or patches