Earlier quoted context omitted.
> Unless you want to scan all chat messages for potential signs of adults engaging in grooming of children too? Well, the point is to scan all messages, period. And then to detect those that come from predators, not adults. How often do parents convince their children to send... revealing pictures? Or to meet them somewhere? How often do parents introduce themselves to their children in messages? You can't seriously…
Sure, but who's reading the conversation to determine whether it "looks suspicious"? A regex? A neural network? Who decides the algorithm, and do you really can believe they won't ever change it to serve other more nefarious purposes like suppressing dissent?
YES. That's the problem. Whoever controls it has that power. We don't want that. That's the argument against ChatControl: "imagine that those who get in power are against you, and imagine what they can do if they abuse this tool".
But saying that "a law enforcement officer may see a false-positive between a parent and their child and I find this insufferable" won't convince many people, IMHO.