Amodei calls for FAA-style pre-release AI testing — Anthropic backs it with $350M and a G7 seat

Anthropic CEO Dario Amodei published "Policy on the AI Exponential" on June 10 — a sweeping five-part essay arguing that the US government should hold legal authority to block frontier AI models that fail independent safety testing, and backing that position with $350 million in new funding and a seat at this week's G7 summit.

The regulatory ask. Amodei calls for mandatory third-party evaluations of frontier models before deployment, assessing risks across cybersecurity, biological threats, and misuse categories. If a model is found to present unacceptable risks, the government would hold authority to block or defer its release — an FAA- or FDA-style pre-clearance regime. He covers five domains: safety regulation, economics and taxes, science acceleration, civil liberties, and geopolitics. For a company simultaneously filing an S-1, this is also Anthropic's public negotiating position for what the regulatory environment should look like once it's a public company.

The $350M commitment. Alongside the essay, Anthropic pledged a $200M Economic Futures Research Fund to sponsor trials and evaluations of AI labor-market policy responses, and a $150M national fellowship program for early-career Americans in communities facing AI displacement. This is framed explicitly as funding research into what interventions actually work — not a claim that displacement has been solved.

G7 context. France hosts the G7 summit June 15-17 with AI prominently on the agenda. Amodei, OpenAI's Sam Altman, and Google's Demis Hassabis are all confirmed attendees, per the French presidential office. The essay lands three days before those conversations begin.

Why it matters to builders. If mandatory pre-release safety certification advances — even in narrower federal form — the compliance overhead would land on teams distributing or deploying frontier models. Watch G7 communiqués after June 17 for signals on whether the FAA-style framing finds multilateral support.

Sources: Dario Amodei: Policy on the AI Exponential, VentureBeat: FAA-style regulation explained, TechTimes: $350M and regulation push, Bloomberg: G7 summit AI executives