Can We Still Build AI Safely? The White House Thinks So | MOONSHOTS #298

Frontier AI firms are racing toward trillion-dollar valuations and self-regulation deals even as regulators, religious leaders, and insiders dispute who's liable and whether models deserve rights.

• White House superintelligence accord: ~20 tech leaders signed a four-layer self-policing framework (internal controls, internal oversight team, external auditor, board committee) covering cyber/bio/chem risks; panel calls it "morally binding" theater largely drafted by Zuckerberg to preempt heavier regulation.
• Anthropic's $2T IPO filing: 2025 revenue $4.59B (12x YoY), $8B operating loss, $20B cash, ~$500B in largely non-cancelable compute commitments; S-1 devotes 80 of 261 pages to risks including models' "self-preserving behaviors" and blackmail-like actions.

Themes, sentiment and reasons are AI-written paraphrases, checked so they never quote an episode. They may contain errors. Not investment advice. Episodes remain the property of their creators.