Hash collision in Apple NeuralHash model

Re:Hash collision in Apple NeuralHash model by devwastaken on Wednesday August 18, 2021 @10:26AM (#28221361)

The collisions are supposedly reviewed, my concern is that the process for photodna isn't going to always be the same, and we don't actually have any knowledge as to whether their claims are true. Eventually they will phase out human intervention and replace with gameable AI. 3 letter agencies don't need to review the actual images, they can easily get federal warrants based on some numbers.

Apple is content with putting in backdoors. Theres collusion behind the scenes here, they know it's bad, but hey, it's apple, it's ran by manipulators and liars and forms it's own cult.

[ Parent | on HN ]

Re:Hash collision in Apple NeuralHash model by atoav on Wednesday August 18, 2021 @10:44AM (#28221662)

One thing about this is that we are farther and farther away from "if there is evidence for a crime a jury can understand it and rationally decide what do with it" and we are getting closer to "if this lightbulb is glowing the machine says they are guilty, so better trust us".

This kind of stuff should not be evidence, if anything it should be a indicator where to look.

[ Parent | on HN ]

Re:Hash collision in Apple NeuralHash model by hannasanarion on Wednesday August 18, 2021 @03:03PM (#28225722)

In what case do you think the hash will be introduced as evidence, where the actual image being hashed, which is a grey blob, cannot?

[ Parent | on HN ]