News for hackers. Stuff that matters.

A comment on Hash collision in Apple NeuralHash model. Parent comment

Re:Hash collision in Apple NeuralHash model by mannerheim on Wednesday August 18, 2021 @11:17AM (#28222317)
> Apple’s method of detecting known CSAM is designed with user privacy in mind. Instead of scanning images in the cloud, the system performs on-device matching using a database of known CSAM image hashes provided by NCMEC and other child-safety organizations. Apple further transforms this database into an unreadable set of hashes, which is securely stored on users’ devices.

https://www.apple.com/child-safety/pdf/CSAM_Detection_Techni...

[ Parent | on HN ]
Re:Hash collision in Apple NeuralHash model by robertoandred on Wednesday August 18, 2021 @12:11PM (#28223306)
Yeah, so how would you know something would collide with the hash?
[ Parent | on HN ]
Re:Hash collision in Apple NeuralHash model by mannerheim on Wednesday August 18, 2021 @12:17PM (#28223413)
Because the hashes are stored on the user's device?
[ Parent | on HN ]
Re:Hash collision in Apple NeuralHash model by robertoandred on Wednesday August 18, 2021 @12:18PM (#28223426)
No, encrypted and blinded hashes are stored. You can’t extract them.
[ Parent | on HN ]
Re:Hash collision in Apple NeuralHash model by fastball on Thursday August 19, 2021 @12:19AM (#28230272)
Right, but your device is checking against the hashes, no?

So at some point you generate an image which triggers a subroutine and you know that image collides.

[ Parent | on HN ]
Re:Hash collision in Apple NeuralHash model by cyanite on Wednesday August 18, 2021 @04:11PM (#28226447)
This also means that the device doesn’t know if a given image is a match.
[ Parent | on HN ]
Re:Hash collision in Apple NeuralHash model by robertoandred on Wednesday August 18, 2021 @04:39PM (#28226799)
Correct, only the server does.
[ Parent | on HN ]