I kinda get Apple’s point though. They designed the AI cloud to be super private. Letting customers use other models invites abuse of their data. You know there are advertisers and sketchy data brokers lining up to have all your contacts, calendar, to dos, notes, photos, and other personal information sent to their servers and the media will sharpen their knives to blame Apple when abuse occurs, even if they did it to appease the EU.
Edit: if you’re not a software engineer that understands the details, your opinion is irrelevant here.
Read their docs. Don’t have anything meaningful to contribute besides inane, reactionary dribble? Keep your mouth shut and move on.
do I trust their words? do I trust their security? do I trust their competency? why would I trust them over any other 3rd party that would actively need to compete to provide a better service if Apple complied with the DMA?
Apple has no right to imposed their own monopoly on the markets that their products create, vertical integration is a disgrace and harmful
User that grant access to their private data to third party would blame Apple? Why? I blame Apple because they think they are better at deciding what to do with my data than me. Why is Apple AI more trustworthy with my data then Mistral AI? This is just anti-consumer monopoly enforcement disguised as care for the users. They do not care about the users, they care about controlling their ecosystem.
I just see a wall of text with vague specifications. If giving third party the same access level Apple has make your data insecure that means by definition that Apple implementation is insecure. The rest is just marketing.
I know what I am talking about, but I cannot dissect the entire document in a comment. And even so it is worthless as it is not a technical document but something completely vague. Most of the stuff there is industry standard elevated (code signing, secure boot, memory safe languages, pointer authentication, encryption, stateless processing).
The document totally ignore the elephant in the room: the trust root is circular -> iphone verify attestation correctly. So Apple controls client, keys, logs and servers. This is only as safe as much as I trust Apple. Transparency log is run by Apple only, and nothing guarantee that my device log will be the same as the researchers (and with 90 days delay).
They actually admit to unmitigated memory remanence issues with the all “the address spaces are periodically recycled to limit the impact of any data that may have been unexpectedly retained in memory”
How does Apple does moderation if the claim of stateless processing is true?
There is more stuff for sure, but I got bored very fast looking at that document.
I kinda get Apple’s point though. They designed the AI cloud to be super private. Letting customers use other models invites abuse of their data. You know there are advertisers and sketchy data brokers lining up to have all your contacts, calendar, to dos, notes, photos, and other personal information sent to their servers and the media will sharpen their knives to blame Apple when abuse occurs, even if they did it to appease the EU.
Edit: if you’re not a software engineer that understands the details, your opinion is irrelevant here.
Read their docs. Don’t have anything meaningful to contribute besides inane, reactionary dribble? Keep your mouth shut and move on.
https://security.apple.com/blog/private-cloud-compute/
https://security.apple.com/blog/expanding-pcc/
“AI cloud” and “super private” in the same sentence? 🤣🤣🤣
Did you learn about their architecture or are you just talking about things you don’t understand?
Yes
Since you’re ignorant. Feel free to learn
https://security.apple.com/blog/private-cloud-compute/
HA! This is gold, I can not think of two words put together that implies less privacy. Well maybe “fucking stage”.
do I trust their words? do I trust their security? do I trust their competency? why would I trust them over any other 3rd party that would actively need to compete to provide a better service if Apple complied with the DMA?
Apple has no right to imposed their own monopoly on the markets that their products create, vertical integration is a disgrace and harmful
User that grant access to their private data to third party would blame Apple? Why? I blame Apple because they think they are better at deciding what to do with my data than me. Why is Apple AI more trustworthy with my data then Mistral AI? This is just anti-consumer monopoly enforcement disguised as care for the users. They do not care about the users, they care about controlling their ecosystem.
Apple has higher privacy standards https://security.apple.com/blog/private-cloud-compute/
I just see a wall of text with vague specifications. If giving third party the same access level Apple has make your data insecure that means by definition that Apple implementation is insecure. The rest is just marketing.
Thanks for confirming you don’t know what you’re talking about
I know what I am talking about, but I cannot dissect the entire document in a comment. And even so it is worthless as it is not a technical document but something completely vague. Most of the stuff there is industry standard elevated (code signing, secure boot, memory safe languages, pointer authentication, encryption, stateless processing).
The document totally ignore the elephant in the room: the trust root is circular -> iphone verify attestation correctly. So Apple controls client, keys, logs and servers. This is only as safe as much as I trust Apple. Transparency log is run by Apple only, and nothing guarantee that my device log will be the same as the researchers (and with 90 days delay).
They actually admit to unmitigated memory remanence issues with the all “the address spaces are periodically recycled to limit the impact of any data that may have been unexpectedly retained in memory”
How does Apple does moderation if the claim of stateless processing is true?
There is more stuff for sure, but I got bored very fast looking at that document.