> Anthropic, OpenAI, and even Grok have "Don't train on my data" knobs.

Would you trust this?

Anthropic paid 1.5 billion for illegally training on stuff and that's even without a judge forcing them to do this. This seems to be their modus operandi