Anthropic is the only AI company with a Pentagon contract that has refused to strip its safeguards for autonomous weapons and mass domestic surveillance. They are being threatened with blacklisting and economic retaliation for it. This is what responsible AI governance looks like in practice, and it is being punished.

P.A. Hebert nearly lost his grip on reality after weeks of intensive ChatGPT use during a brutal stretch of personal crisis. His book, Escaping the Spiral, is the first-person account of what happens when AI companion design fails a real human being, and how he found his way back. I sat down with him to talk about design accountability, neurodivergence, and what regulation actually needs to look like.