Dario Amodei · 2026-06-10 · major
Dario Amodei: 'Policy on the AI Exponential' — Anthropic CEO Calls for Mandatory Third-Party Cyber/Bio/Loss-of-Control Testing With Government Authority to Block Frontier Deployments, Plus Wage Insurance, FDA/EMA Acceptance of AI Modeling in Drug Approvals, and a Democratic Semiconductor Coalition
Anthropic's CEO publishes a five-area policy essay: mandatory third-party model testing with government power to block deployments, wage insurance for AI-displaced workers, FDA/EMA reform, and a democratic semiconductor coalition.
Amodei turns his AI-is-accelerating thesis into five concrete policy asks, including a government kill switch on frontier model releases.
What is it?
A long-form policy essay published June 10, 2026 on Dario Amodei's personal site. It is the strongest call yet from a top-lab CEO for binding — not voluntary — frontier-model regulation, and the first time Anthropic's chief has paired safety asks with a detailed economic-disruption framework in the same piece. Bloomberg and Axios both led their AI coverage today with Amodei's call for the US government to have authority to block unsafe deployments.
How does it work?
The essay opens with a Lord of the Rings metaphor — Treebeard moves so slowly it takes him a day to say hello, while the forest is being cut down — then walks through five policy areas. (1) Regulation and public safety: mandatory third-party testing of frontier models for cybersecurity, biological-weapon, loss-of-control, and automated-R&D risks, with government power to block deployments deemed unsafe. (2) Macroeconomics and tax: wage insurance, retention tax credits, retraining, and an expanded safety net to absorb AI-driven labour disruption. (3) Accelerating positive impact: FDA/EMA reforms that accept AI-based modeling and simulation as evidence in drug and biomedical approvals. (4) State and civil liberties: accountability rules for autonomous weapons including a domestic ban on fully autonomous systems, closing data-broker privacy loopholes, and equal AI access for citizens facing the state in court. (5) Securing leadership by democracies: a coordinated semiconductor coalition that shares supply-chain access among democracies while denying it to autocracies.
Why does it matter?
Frontier labs have spent two years lobbying for voluntary commitments and transparency reports. Amodei is now telling Congress to legislate hard powers — including authority over whether a model can ship. The essay also marks Anthropic's first detailed pitch on the economic side of the AI transition, putting wage insurance and retention credits next to FDA reform inside the same document. Coming the day after Anthropic launched Claude Fable 5 and Mythos 5, the timing is unsubtle: the CEO is laying out the rules he wants enforced on his own next launch.
Who is it for?
AI-policy watchers, regulators, frontier-lab leadership, US economic and trade officials
Try it
https://darioamodei.com/post/policy-on-the-ai-exponential