If this has the effect its proponents claim it will, it seems like substantially the larger outcome of this will be that government agents will be reviewing people's sexts. They say that false positives are rare, but how often is it okay for the government to be reviewing peoples' sexts? I found it a little hard to get concrete info on how exactly their image hashes work, but it sounds quite literally that if you've got a couple of young people (whether teens or twentysomethings) who are sexting, and their sexts look a bit like some piece of "known CSAM" if you squint, then a government agent will review it and possibly harass them.
Seems like eventually the law will get some poor girl killed when the authorities contact her parents about "CSAM," discover that it was the girl herself who took the picture and sent it to her boyfriend, her dad finds out she was having sex and does an honor killing.
But we're just supposed to trust that these image hashes have a small false positive rate, when there's no way to have transparent review without making it easy for adversaries to avoid the scan.
Even if they have a small false positive rate, the absolute figures will be staggering. 500 million people and all texts are being scanned, with more than 99% not being CSAM.. you do the maths..
Seems like eventually the law will get some poor girl killed when the authorities contact her parents about "CSAM," discover that it was the girl herself who took the picture and sent it to her boyfriend, her dad finds out she was having sex and does an honor killing.
But we're just supposed to trust that these image hashes have a small false positive rate, when there's no way to have transparent review without making it easy for adversaries to avoid the scan.