4 ms·

My opinion of Apple falls pretty substantially with every new justification and explanation. They invented a fancy back door. Privacy-preserving software cannot

My opinion of Apple falls pretty substantially with every new justification and explanation. They invented a fancy back door. Privacy-preserving software cannot trigger third party content review under any circumstances, no matter how vile. It cannot fail in a way that results in an arrest or legal expenses. If Apple won’t indemnify customers against false positives produced by this system, then it isn’t reliable enough for me to entrust them with my data.

alwillis 5y ago
They just revealed that the threshold is around 30 CSAM images that would result in Apple knowing anything about your device. And even then, humans at Apple do a review to see what’s happening; if it’s not CSAM, nothing else happens. It doesn’t seem that false positives are possible—either it’s CSAM from the database or it’s not. But even in the event of some unforeseen edge case, humans at Apple check things out before anything else can happen. But that would have to be quite the edge case to meet the threshold of around 30 possible CSAM images before Apple can know anything about a user.
- JohnFen 5y ago
  All of which is irrelevant to what I consider the largest problem with this scheme: that it's being done on-device instead of on the server. This is little different than putting surveillance equipment in people's homes to see what they're putting into boxes that are going to be be mailed, as opposed to examining the contents of the boxes at the post office. In either case, the surveillance itself is equally problematic, but in the case of doing it in your home, a dangerous precedent has been set that makes it easier to implement other, perhaps more questionable, home surveillance.
  - alwillis 5y ago
    All of which is irrelevant to what I consider the largest problem with this scheme: that it's being done on-device instead of on the server. There are components that run on the device and the server. One can't function without the other. That's why not using iCloud Photos disables the entire thing.

alwillis 5y ago

They just revealed that the threshold is around 30 CSAM images that would result in Apple knowing anything about your device. And even then, humans at Apple do a review to see what’s happening; if it’s not CSAM, nothing else happens. It doesn’t seem that false positives are possible—either it’s CSAM from the database or it’s not. But even in the event of some unforeseen edge case, humans at Apple check things out before anything else can happen. But that would have to be quite the edge case to meet the threshold of around 30 possible CSAM images before Apple can know anything about a user.

JohnFen 5y ago
All of which is irrelevant to what I consider the largest problem with this scheme: that it's being done on-device instead of on the server. This is little different than putting surveillance equipment in people's homes to see what they're putting into boxes that are going to be be mailed, as opposed to examining the contents of the boxes at the post office. In either case, the surveillance itself is equally problematic, but in the case of doing it in your home, a dangerous precedent has been set that makes it easier to implement other, perhaps more questionable, home surveillance.
- alwillis 5y ago
  All of which is irrelevant to what I consider the largest problem with this scheme: that it's being done on-device instead of on the server. There are components that run on the device and the server. One can't function without the other. That's why not using iCloud Photos disables the entire thing.