Anthropic bets on a safety arms race across states

Anthropic is pushing a state‑by‑state strategy to raise AI safety standards rather than push for a single national framework, contrasting with OpenAI’s effort to mirror laws across states. CEO Dario Amodei’s former company is backing tougher protections for the most capable models, citing California’s 2025 law as a springboard and supporting stricter measures in New York, Illinois, and Massachusetts—including independent audits and enforcement powers. The approach unfolds as federal action stalls, with both firms expanding political activity as they jockey to shape how U.S. AI safety rules take shape.