3 ms·

The police will only be involved after their scanner has found 30 matches and Apple’s reviewer has manually concluded that it actually is CSAM, so I don’t see h

The police will only be involved after their scanner has found 30 matches and Apple’s reviewer has manually concluded that it actually is CSAM, so I don’t see how the police is really relevant? If HOA means home owners association, I don’t see the connection. They can’t influence Apple either.

ClumsyPilot 5y ago
Its the thin end of the wedge argument, and the fact that your property 'the phone' is spying on you and reporting you to the police.
heavyset_go 5y ago
> * Apple’s reviewer has manually concluded that it actually is CSAM* There is no manual review of the image by Apple employees. What is viewed is a derivative image, which is scaled down and blurred. It would be trivial for someone to use legal pornographic images and put it through a NeuralHash collision generator, which upon viewing the scaled down derivative image, would look like CSAM.
- Grustaf 5y ago
  If that’s what you’re worried about, then if people have physical control over your phone they could just upload CSAM to your FB or Google photos from it.
  - brendoelfrendo 5y ago
    There's a world of difference between someone creating a NeuralHash collision and spamming people's iMessage with false positives and actually taking the device and uploading CSAM to the cloud.
    - Grustaf 5y ago
      Well yes, spamming your iMessages would be pointless.
  - heavyset_go 5y ago
    I said nothing about physical control. WhatsApp and other apps automatically download images that are sent to them. Scammers and those who commit fraud or phishing attempts don't have physical access to their victims electronic, yet they're able to get victims to download and run malicious code all of the time. There's also the fact that iOS exploits are so plentiful that they're cheaper than Android exploits.
    - Grustaf 5y ago
      Still, why not send actual CSAM then, so that they are actually caught? Getting them into review just causes them inconvenience. Remember, only the flagged images will be reviewed, and only they could ever be sent to law enforcement, so it wouldn't even reveal their private photos or anything like that.
      - heavyset_go 5y ago
        > Still, why not send actual CSAM then, so that they are actually caught? Getting them into review just causes them inconvenience. Because that requires possessing illegal images, while legal pornography that also happens to cause a hash collision with NeuralHash doesn't. Again, no one at Apple is reviewing the source image. They're reviewing obfuscated derivative images, and then informing law enforcement if they suspect they're illegal images. What law enforcement then does is get warrants for all of the accused's electronic devices and raids their home and workplace in order to collect evidence. It's another version of SWATing. Sure, eventually the authorities might find out that the accused isn't actually a mass shooter or holding hostages, but by then the damage is already done.
        Grustaf 5y ago
        Why would the police raid you based on blurry versions of completely legal images? And are you really saying that the images Apple reviews are so blurry that they can’t review them? That seems like a very stupid system. In any case, your scenario sounds pretty far fetched. Even if it actually works, I wouldn’t say it’s a major blocker.

ClumsyPilot 5y ago

Its the thin end of the wedge argument, and the fact that your property 'the phone' is spying on you and reporting you to the police.

heavyset_go 5y ago

> * Apple’s reviewer has manually concluded that it actually is CSAM* There is no manual review of the image by Apple employees. What is viewed is a derivative image, which is scaled down and blurred. It would be trivial for someone to use legal pornographic images and put it through a NeuralHash collision generator, which upon viewing the scaled down derivative image, would look like CSAM.

Grustaf 5y ago
If that’s what you’re worried about, then if people have physical control over your phone they could just upload CSAM to your FB or Google photos from it.
- brendoelfrendo 5y ago
  There's a world of difference between someone creating a NeuralHash collision and spamming people's iMessage with false positives and actually taking the device and uploading CSAM to the cloud.
  - Grustaf 5y ago
    Well yes, spamming your iMessages would be pointless.
- heavyset_go 5y ago
  I said nothing about physical control. WhatsApp and other apps automatically download images that are sent to them. Scammers and those who commit fraud or phishing attempts don't have physical access to their victims electronic, yet they're able to get victims to download and run malicious code all of the time. There's also the fact that iOS exploits are so plentiful that they're cheaper than Android exploits.
  - Grustaf 5y ago
    Still, why not send actual CSAM then, so that they are actually caught? Getting them into review just causes them inconvenience. Remember, only the flagged images will be reviewed, and only they could ever be sent to law enforcement, so it wouldn't even reveal their private photos or anything like that.
    - heavyset_go 5y ago
      > Still, why not send actual CSAM then, so that they are actually caught? Getting them into review just causes them inconvenience. Because that requires possessing illegal images, while legal pornography that also happens to cause a hash collision with NeuralHash doesn't. Again, no one at Apple is reviewing the source image. They're reviewing obfuscated derivative images, and then informing law enforcement if they suspect they're illegal images. What law enforcement then does is get warrants for all of the accused's electronic devices and raids their home and workplace in order to collect evidence. It's another version of SWATing. Sure, eventually the authorities might find out that the accused isn't actually a mass shooter or holding hostages, but by then the damage is already done.
      - Grustaf 5y ago
        Why would the police raid you based on blurry versions of completely legal images? And are you really saying that the images Apple reviews are so blurry that they can’t review them? That seems like a very stupid system. In any case, your scenario sounds pretty far fetched. Even if it actually works, I wouldn’t say it’s a major blocker.