Hash collision in Apple NeuralHash model

Re:Hash collision in Apple NeuralHash model by IncRnd on Wednesday August 18, 2021 @10:11PM (#28229572)

> This isn’t true at all.

> If the creation of fakes is as easy as claimed, Neuralhash evidence alone will become inadmissible.

Okay. https://github.com/anishathalye/neural-hash-collider

[ Parent | on HN ]

Re:Hash collision in Apple NeuralHash model by spurgu on Wednesday August 18, 2021 @11:14PM (#28229946)

Uh? So his if statement is true?

[ Parent | on HN ]

Re:Hash collision in Apple NeuralHash model by IncRnd on Wednesday August 18, 2021 @11:21PM (#28229981)

Please read what is written right before that... You are taking something out of context.

[ Parent | on HN ]

Re:Hash collision in Apple NeuralHash model by zepto on Saturday August 21, 2021 @10:19PM (#28262326)

Why do you keep posting links to this collider as though it means something?

As has been already pointed out the system is designed to handle attacks like this.

Here is the relevant paragraph from Apple’s documentation:

“as an additional safeguard, the visual derivatives themselves are matched to the known CSAM database by a second, independent perceptual hash. This independent hash is chosen to reject the unlikely possi- bility that the match threshold was exceeded due to non-CSAM images that were ad- versarially perturbed to cause false NeuralHash matches against the on-device en- crypted CSAM database. If the CSAM finding is confirmed by this independent hash, the visual derivatives are provided to Apple human reviewers for final confirmation.”

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

[ Parent | on HN ]