Hacker News new | ask | show | jobs
by hypothesis 1766 days ago
They will be if you collide a low-res image that resembles CSAM.

Why would person doing manual review risk his job in case if he’s unsure? Naturally he will just play it safe and report images.

1 comments

Not resembles. The adversarial image has to match a private perceptual hash function of the same CSAM image that the NeuralHash function matched before a human reviewer ever looks at it.
Do you have any material on this private function?
Not beyond the documents Apple has shared. Presumably it will be kept that way given it prevents an adversarial attack against it.