So what? If predators cannot talk to children over SnapChat, that's a win, wouldn't you say?

The only valid argument I see against ChatControl is that fundamentally, you cannot know what it is reporting. It's not like if there would be an open source list of illegal material together with the hashes, right?

If you cannot audit what is being reporting (with whatever means necessary to make sure it is doing what it should be doing), then whoever controls it could abuse it.

That's the problem. That's the reason not to implement it. But it's completely overwhelmed by the flood of invalid arguments.