Peter Thiel’s Antichrist lectures call AI safety the road to control, while his fund backs OpenAI and Anthropic. His own 2004 essay explains why.