Zvi: Anthropic Doesn't Understand Why Its Draft AI Bill Was Unacceptable
Don't Worry About the Vase (Zvi) · rss · 2026-10-21
Zvi Mowshowitz dissects Anthropic's leaked draft federal AI safety bill and Anthony Cimino's official response, arguing Anthropic addresses marginal concerns without grasping what was centrally wrong. On preemption, the promised 'very strong federal standard' isn't. On model blocking, Anthropic's occupational-safety analogy fails because frontier AI risks can be irreversible — blocking must be 'take it down first, argue later,' especially since the lab-picked evaluator may be the government's only eyes. On risk thresholds, the Hugging Fable case shows nothing triggers without $1B in damages or 50 deaths. Circulating the bill as an anchor in September signals Anthropic is fine with compromises it shouldn't accept — the balance of power is shifting.
More from Companies & People
- YC cofounder after NYC coffee chats: optimism is a way-underrated founder quality — ycombinator · 2026-10-21
- YC and Cotool panel: how hyper-growth startups build security orgs from zero with AI agents — ycombinator · 2026-10-21
- Anthropic researcher debunks 'Claude is uniquely condescending' claims in viral spat — BlancheMinerva · 2026-10-21
- Fireworks, Microsoft and NVIDIA host SF buildathon: beat a frontier model, win $10K — sophiamyang · 2026-10-21
- Engineer slams Gartner as a 'tax' enterprises pay to protect jobs and block innovation — matt_slotnick · 2026-10-21
- FirstMark portfolio in one day: Daytona Series B, Baba Series A, Gradium's on-device voice model — mattturck · 2026-10-21