Eventually, yes. As we work more towards AI's and "smart" image recognition eventually there will be a system that will have these false positives on innocent i

Eventually, yes. As we work more towards AI's and "smart" image recognition eventually there will be a system that will have these false positives on innocent images. Current CSAM depends on humans to "verify" the imagery, this is something companies desperately want to get rid of, and so do the employees understandbly so. Nobody wants a job (99.99%) of comparing CSAM. It costs companies money in labor costs, and draws them bad PR when those employees inevitably develop permanent/semi-permanent mental health issues from it. The only reason it hasn't happened yet is because a startup can't just start scanning CSAM. They need the blessing of the feds to do that, which requires political connections, and of course requires competing with companies that already have that blessing - something that politics prevents. PhotoDNA and current CSAM scanning only gets known CSAM, but not new CSAM. The end goal is to detect CSAM before it's ever even distributed, to be "closer to the victim", rather than just those consuming it. Even with current PhotoDNA you can generate hash collisions, which flag the image for review, and a real human compares the material. This is of course subject to change for the reasons stated above. Secondarily, automatic scanning and ID'ing of imagery is how you can easily throw an FBI raid at someone. Apps like Telegram automatically download every image/video in the thread. Ontop of that you can create images that appear different at differing resolutions. At one resolution a harmless meme, at another, CSAM. Meaning that you can again throw an FBI raid at someone using simple tricks.