I believe everyone who can should try this. Be warned: Meta and Bytedance don't do well at rate limiting. Anthropic and OpenAI do better, but will still place appreciable load on web servers.

Residential proxy scrapers will fill in for all of these somewhat unethical companies if they figure it out, I bet.

I think everyone is feeding them fake data, and that's the problem. They're basically poisoned search engines. As are many people these days.