Hacker News Reader

Comment by nojito | Hacker News Reader

Why are hash collisions relevent?

There are atleast 2-3 further checks to account for this.

Because it's not a cryptographic hash where a one bit difference results in a completely different hash. It's a perceptual hash that operates on a smaller bitmap derived from the image so it's plausible that some innocuous images might result in similar derivations; and there might be intentionally crafted innocently-looking images that result in an offensive derivative.

Salvador Dali could do something similar by hand in 1973 in Gala Contemplating the Mediterranean Sea [1]

[1] https://en.wikipedia.org/wiki/Lincoln_in_Dalivision

This is a great answer but that’s not actually the GP’s contention. Their argument is essentially “so what if there’s a collision, the human review will catch it”. And to that I’d say that the same is supposed to occur for the no-fly list and we all know how that works in practice.

The mere accusal itself of possessing CSAM can be life ruining if it gets to that stage. More importantly, a collision will effectively allow warrantless searches, at least of the collided images.

Indeed, I touched on that in another comment: https://news.ycombinator.com/item?id=28227141

That’s one check. There are other system checks to make the client side hash collision meaningless.

Do you understand that anyone can take absolutely legal porn and make it match CSAM hash? And no one except NCMEC can know the difference because they all only compare hashes and not actual images.

And whoever going to check images for Apple will see that yeah, there is porn on picture. Flag it. Then you'll have unlimited amount of time to explain to FBI why some porn on your device match CSAM hash.

Reply on news.ycombinator.com