[HN Gopher] Apple will soon support encrypted RCS messaging with...
       ___________________________________________________________________
        
       Apple will soon support encrypted RCS messaging with Android users
        
       Author : 01-_-
       Score  : 64 points
       Date   : 2025-03-14 12:07 UTC (10 hours ago)
        
 (HTM) web link (www.theverge.com)
 (TXT) w3m dump (www.theverge.com)
        
       | thomascountz wrote:
       | Announcement discussion from 2023:
       | 
       | Apple announces that RCS support is coming to iPhone next year
       | 
       | https://news.ycombinator.com/item?id=38293082
       | 
       | 793 points | 709 comments
        
         | pityJuke wrote:
         | To be clear, this is a different announcement. That was just
         | RCS, this is encrypted RCS.
        
         | cantSpellSober wrote:
         | The big difference:
         | 
         | > _The GSM Association announced that the latest RCS standard
         | includes E2EE_
         | 
         | > _While Apple's proprietary iMessage system already supported
         | E2EE, this wasn't extended to RCS messaging because the
         | previous RCS standard didn't provide cross-platform support_
        
       | criddell wrote:
       | When an iPhone sends a message to 555-1212, where does the iPhone
       | get the public key for that number?
        
         | acdha wrote:
         | Apple runs servers for managing iMessage key exchange, just
         | like Google's RCS encryption. Both of them use device
         | attestation to restrict access to those servers to their own
         | apps.
        
       | oneplane wrote:
       | So essentially, it's not really an Apple thing, it's more that
       | the universal RCS profile just didn't have encryption, and Google
       | RCS was a non-standard extension that nobody else was allowed to
       | use.
       | 
       | The real news is the update to GSMA RCS, because without that,
       | none of this matters. What I'm missing in the article is who's
       | going to own the keys and why this is probably going to default
       | to the telcos as if it's MMS. Are they going back to the days of
       | charging per message?
       | 
       | With iMessage you'd be putting your trust in Apple, with Google
       | RCS you'd be putting your trust in Google. For WhatsApp that'd be
       | Meta and for Signal that's Signal. But with GSMA RCS?
        
         | she46BiOmUerPVj wrote:
         | Is there anything that shows "no one was allowed to use it" or
         | was it that it wasn't an accepted standard?
        
           | oneplane wrote:
           | Yes, Google only made it available for Google Messages. They
           | don't have an SDK or API you can use to make your own clients
           | or servers. Google also didn't put it up for standards with
           | the GSMA or any other standards body, at least not publicly.
           | There are no records of it.
           | 
           | There are some older submissions here you can probably find
           | using one of the HN search sites about this, but IIRC those
           | didn't really have any internal Google policy about this,
           | they kept it all pretty private. The only 'leak' I remember
           | about this was the thing where manufacturers that preload
           | Google Android have to ship Google Messages to get RCS
           | support from Google, otherwise they can't have it. Also means
           | you can't have RCS without Play Services.
        
             | jauntywundrkind wrote:
             | The destruction of third party apps has been totally wild.
             | 
             | I'm very curious how long this OS-coupled status quo is
             | going to go on for.
        
           | acdha wrote:
           | Look at Google's documentation: they explicitly state that
           | only their Messages app is allowed to talk to their key
           | exchange server. The entire marketing campaign they ran about
           | RCS was predicated on nobody reading their docs or noticing
           | all of the Android developers begging for permission to use
           | RCS for years.
           | 
           | https://www.gstatic.com/messages/papers/messages_e2ee.pdf
           | 
           | > E2EE is implemented in the Messages client, so both clients
           | in a conversation must use Messages, otherwise the
           | conversation becomes unencrypted RCS. In rare situations
           | where the conversation starts as E2EE, then one of the
           | clients migrates to a different RCS client or an older
           | Messages client that does not support E2EE, Messages might be
           | unable to detect the change immediately. If the Messages user
           | sends a new message, it's still E2EE, however the recipient
           | client may render the encrypted base64 payload directly as
           | message content.
        
         | stephenr wrote:
         | I understand your point that the "news" part is that RCS
         | standard now includes E2EE, rather than about Apple's support
         | for said standard.
         | 
         | But I don't think it's fair to suggest or imply that this
         | development is unrelated to Apple either.
         | 
         | RCS has been a thing for nearly a decade, and _Google_ 's RCS
         | backend has been doing non-standard E2EE for half that time.
         | 
         | Within 8 months of Apple publicly announcing they would adopt
         | RCS and work with GSMA to support _standardised_ E2EE, there is
         | suddenly a standard for it...
        
         | jeroenhd wrote:
         | The encryption is based on MLS: https://www.rfc-
         | editor.org/rfc/rfc9420.html
         | 
         | I don't think Google wanted to gatekeep their E2EE
         | implementation. They have some generic documentation about how
         | it works:
         | https://www.gstatic.com/messages/papers/messages_e2ee.pdf
         | 
         | The thing about RCS is that no messengers seem to care at all
         | about implementing RCS themselves. Part of that is probably
         | because depending on the carrier, RCS may require access to
         | certain SIM card information, which only pre-installed apps can
         | do, and part of it is that many developers are waiting for
         | Google to add RCS to the same API that SMS/MMS already exposes
         | because they don't want to implement RCS themselves.
         | 
         | Realistically, the target demographic for their documentation
         | is 1) Apple (who they'd happily supply with details to get rid
         | of the green bubble problem) and maybe 2) government officials
         | looking into antitrust concerns. In theory someone working for
         | LineageOS can implement an RCS client, though, but for those
         | developers I don't think reverse engineering the remaining
         | unknowns about the protocol (mostly "what server" and "what
         | message contents") aren't that difficult.
         | 
         | I'm not cryptographer, but I haven't heard any major issues
         | from actual cryptographers about MLS. It's encryption
         | principles seem to be similar to those of Signal. Google is
         | actually already using MLS in their proprietary E2EE
         | implementation.
         | 
         | Ideally, MLS would be combined with MIMI so that messaging apps
         | become interoperable, but that's probably a pipe dream.
        
           | acdha wrote:
           | The protocol is open but key exchange is not: even on
           | Android, third-party messengers can't interoperate with
           | Google Messages. See page 11 of that PDF.
        
       | mrweasel wrote:
       | Why was RCS even designed with a none encrypted mode? I get that
       | the original spec isn't exactly new, but it's also not so old
       | that encryption, security or privacy wasn't an issue.
        
         | nubinetwork wrote:
         | I'm not sure about the case of RCS, but I've seen some
         | instances of a none cipher being better for compression and
         | deduplication, because the encryption messes with the data.
        
           | hagbard_c wrote:
           | You'd normally compress the data before encrypting it as that
           | makes the resulting cyphertext more resilient against
           | cryptanalysis as well as reduces the amount of data which
           | needs to be encrypted so this sounds like a bogus reason.
        
             | nubinetwork wrote:
             | It depends on what you're doing I guess... for storage it
             | doesn't make sense, but for pushing a lot of data over a
             | private pipe, why spend the resources adding a cipher?
        
             | jayd16 wrote:
             | You'd have to compress every single message separately and
             | then encrypt them. That's still a far cry from being able
             | to compress across every message.
        
           | dontlaugh wrote:
           | I assume some sort of multicast.
        
         | jeroenhd wrote:
         | It's the evolution of SMS/MMS; development started 18 years
         | ago, in 2007. The modern spec is based on that with a whole
         | bunch of additional revisions for things like video calling and
         | transferring money. It was designed long before major messenger
         | apps had e2ee in the first place.
         | 
         | Had it been designed with the security practices at the time,
         | the protocol would've been ossified to the point of being
         | practically insecure by today's standards. In a sense, the fact
         | nobody cared about it until the spec was old enough to drive is
         | actually good for users.
         | 
         | The GSMA which designs RCS also serves the needs of government
         | agencies that are tracking (international) criminals, so I bet
         | there must have been some pretty strong opposition against E2EE
         | in the official spec. Frankly, I'm surprised they're even
         | putting it in the spec.
        
         | izacus wrote:
         | Encryption and privacy isn't an issue for carriers, which were
         | part of the standard body and operate SMS protocol even in
         | 2025.
         | 
         | You assume everyone has the same goals in mind :)
        
       | snowwrestler wrote:
       | Now if only we could get Google to support RCS in Google Voice.
        
         | toomuchtodo wrote:
         | It's really odd they don't, I'm incredibly curious why.
        
           | chneu wrote:
           | GVoice gets forgotten about, I swear.
           | 
           | There are years between updates and then suddenly for like 6
           | months it'll get a slew of updates, most of which aren't
           | user-facing. Then back to update silence. It's really odd.
           | 
           | I'm still amazed that google hasn't killed it.
        
             | the-rc wrote:
             | I always wondered if it's still alive only because Larry or
             | Sergey use it every day.
        
       | codedokode wrote:
       | It is good that commercial messengers forced GSM association to
       | finally create E2EE standards. The reason why telcos want you to
       | use RCS becomes obvious if you calculate how much 1Gb of data
       | costs if sent as SMS (I guess that one could buy a car with this
       | money).
        
       | qrush wrote:
       | Still waiting for Google Fi to turn RCS on for iOS users :(
        
         | bilal4hmed wrote:
         | its coming
         | 
         | https://9to5google.com/2025/03/04/google-fi-rcs-ios-18-4/
        
       | ebb_earl_co wrote:
       | This is great in principle. I still prefer Signal as the top-
       | shelf experience of iOS--Android communication.
        
       | tengbretson wrote:
       | Still waiting on a non-Google implementation of this so-called
       | standard.
        
       | WorldPeas wrote:
       | Any news on if/when one will be able to send RCS from a program
       | or device outside the monolith?
        
         | bebna wrote:
         | Never. It is just a fancy MMS in new disguise. Carriers /
         | patent holders will hold to it forever while praising it as the
         | new best thing.
        
           | singpolyma3 wrote:
           | You can get access to send MMS though. RCS there is no one
           | selling access yet, especially for person to person.
        
       | alwayslikethis wrote:
       | Unfortunately, RCS on Android requires google apps, so this isn't
       | really a solution to anyone who doesn't want to be tracked by
       | Google everywhere they go.
       | 
       | I'm still a little confused as to what problem RCS is supposed to
       | solve. It is just as centralized as any other chat app, and is a
       | bit more invasive (often requiring device attestation). Is it
       | really worth all this hassle just to not have to install, let's
       | say, Signal?
        
       ___________________________________________________________________
       (page generated 2025-03-14 23:01 UTC)