Anthropic's IPO filing reveals the tension between safety pledges and rule-breaking revenue

TheTuringPost · x · 2026-11-19

Turing Post #173 takes a deep look at Anthropic's IPO filing: the prospectus shows how much revenue came from customers later found to have broken its rules, and why privacy agreements can make misuse harder to investigate — leading the author to argue "safety in AI is fake."

Other highlights: two people behind Claude Code, Thariq Shihipar and Anthony Morris, announced departures right before the filing; researcher Jacob Coxon left before his equity vested, warning about the speed of the AI race; safety team member Joe Benton joined an independent evaluator. The issue also covers Block's Dessa learning from human conversation and robots learning from shared experience.

Original post →

More from Companies & People

Companies & People channel →