5 ms·

can you explain? Apple have said that if you have x amount of hash matches/ collisions to a set of images you'll be flagged to the authorities? This site has

can you explain? Apple have said that if you have x amount of hash matches/ collisions to a set of images you'll be flagged to the authorities? This site has proven now that it's easy to create pictures whose hash matches (colides!) thus potentailly flagging lots of poeple

judge2020 5y ago
After 30 hash matches/collisions[0], Apple's system is cryptographically[1] able to open the collisions and only then is it sent to Apple's review team for review. If the manual review team confirms the CSAM, it's reported to NCMEC and then NCMEC's team will likely file a police report if they also confirm the images are CSAM. 0: https://pocketnow.com/neuralhash-code-found-in-ios-14-3-apple-not-the-one-well-use-for-csam-photo-scanning#:~:text=Apple%20also%20said%20that%20after%20a%20user%20passes%20the%2030%20match%20threshold%2C%20a%20second%20non-public%20algorithm%20that%20runs%20on%20Apple%E2%80%99s%20servers%20will%20check%20the%20results https://pocketnow.com/neuralhash-code-found-in-ios-14-3-appl... 1: https://www.apple.com/child-safety/pdf/CSAM_Detection_Technical_Summary.pdf#page=11 https://www.apple.com/child-safety/pdf/CSAM_Detection_Techni...
zepto 5y ago
https://news.ycombinator.com/item?id=28305946 https://news.ycombinator.com/item?id=28305946
zimpenfish 5y ago
> This site has proven now that it's easy to create pictures whose hash matches (colides!) But no-one, yet, I think, has cracked open the blob containing the neuralhashes of the NCMEC CSAM corpus which are the actual hashes you need to create pictures to collide with to flag people under Apple's new system, right? (This may be because the blob isn't yet on the phones; I don't know, I just haven't seen anyone mention it wrt creating hashes.)
- read_if_gay_ 5y ago
  You just need to get ahold of some widely spread CSAM. Certainly not everybody is ready to do that, but it is far from impossible for a motivated individual to obtain "hot" hashes.
  - vageli 5y ago
    And once created, the list could be distributed ad infinitum. This is "security by obscurity" at its finest (in terms of the lack of availability of the hash database). If it were a truly robust system, why not make the hash database public?
  - zepto 5y ago
    This is false. There is a second independent hash that you don’t have access to that also needs to match.
    - read_if_gay_ 5y ago
      Ever hear about Kerckhoff’s principle?
  - zimpenfish 5y ago
    > You just need to get ahold of some widely spread CSAM. Risky. But ok, you still need to get hold of 30 images[1] that are in the NCMEC corpus before it'll even get flagged up to a human reviewer. [1] Do we know if it's 30 distinct images or will 30 copies of the same one trigger things? The latter is far easier than the former for the gathering step here, of course.
- kristjansson 5y ago
  And there's no preimage attack! So even with the unblinded hashes (which per Apple's description are encrypted with a private key they control) one couldn't construct an innocent image to match a given hash - they would need the original image too!

judge2020 5y ago

After 30 hash matches/collisions[0], Apple's system is cryptographically[1] able to open the collisions and only then is it sent to Apple's review team for review. If the manual review team confirms the CSAM, it's reported to NCMEC and then NCMEC's team will likely file a police report if they also confirm the images are CSAM. 0: https://pocketnow.com/neuralhash-code-found-in-ios-14-3-apple-not-the-one-well-use-for-csam-photo-scanning#:~:text=Apple%20also%20said%20that%20after%20a%20user%20passes%20the%2030%20match%20threshold%2C%20a%20second%20non-public%20algorithm%20that%20runs%20on%20Apple%E2%80%99s%20servers%20will%20check%20the%20results https://pocketnow.com/neuralhash-code-found-in-ios-14-3-appl... 1: https://www.apple.com/child-safety/pdf/CSAM_Detection_Technical_Summary.pdf#page=11 https://www.apple.com/child-safety/pdf/CSAM_Detection_Techni...

zepto 5y ago

https://news.ycombinator.com/item?id=28305946 https://news.ycombinator.com/item?id=28305946

zimpenfish 5y ago

> This site has proven now that it's easy to create pictures whose hash matches (colides!) But no-one, yet, I think, has cracked open the blob containing the neuralhashes of the NCMEC CSAM corpus which are the actual hashes you need to create pictures to collide with to flag people under Apple's new system, right? (This may be because the blob isn't yet on the phones; I don't know, I just haven't seen anyone mention it wrt creating hashes.)

read_if_gay_ 5y ago
You just need to get ahold of some widely spread CSAM. Certainly not everybody is ready to do that, but it is far from impossible for a motivated individual to obtain "hot" hashes.
- vageli 5y ago
  And once created, the list could be distributed ad infinitum. This is "security by obscurity" at its finest (in terms of the lack of availability of the hash database). If it were a truly robust system, why not make the hash database public?
- zepto 5y ago
  This is false. There is a second independent hash that you don’t have access to that also needs to match.
  - read_if_gay_ 5y ago
    Ever hear about Kerckhoff’s principle?
- zimpenfish 5y ago
  > You just need to get ahold of some widely spread CSAM. Risky. But ok, you still need to get hold of 30 images[1] that are in the NCMEC corpus before it'll even get flagged up to a human reviewer. [1] Do we know if it's 30 distinct images or will 30 copies of the same one trigger things? The latter is far easier than the former for the gathering step here, of course.
kristjansson 5y ago
And there's no preimage attack! So even with the unblinded hashes (which per Apple's description are encrypted with a private key they control) one couldn't construct an innocent image to match a given hash - they would need the original image too!