Dario Amodei Wants FAA-Style AI Regulation and 350 Million Dollars for Displaced Workers - Is This the Right Policy

[GoldbergFan]

The Policy on the AI Exponential essay that Dario Amodei published on June 10th is the most substantive policy document to come out of any major AI lab and it deserves to be taken seriously on its merits. Amodei argues for a shift from transparency legislation to binding enforceable regulation, specifically a system where frontier AI models would require independent safety testing before public release, modelled on how the FAA regulates aircraft before they can fly. He also commits 200 million dollars to an Economic Futures Research Fund and 150 million dollars to a national fellowship programme for early-career Americans affected by AI-driven job displacement.

The essay covers five areas: safety regulation, economics and taxes, science acceleration, civil liberties, and geopolitics. The safety regulation section is the most consequential because it explicitly proposes that the US government should have legal authority to block or reverse the release of frontier AI models that fail independent safety testing. This goes substantially further than anything currently under serious consideration in Washington, which as Axios reported is precisely the point Amodei is making. He frames the policy-AI gap using the image of hobbits trying to rouse Treebeard, a wise but ponderous entity that operates at a completely different timescale from the urgency of the situation.

Do you think FAA-style pre-release testing for AI models is actually workable and desirable, and is the 350 million dollar economic commitment meaningful or mostly optics?

[RayOfLight31]

The FAA analogy is better than most AI regulation analogies because aircraft certification is actually about systematic testing before deployment rather than post-hoc liability. That is the right model for AI safety

[FadedKernel]

The problem with FAA-style testing is that we do not yet have a settled definition of what a frontier AI model should be tested against. The FAA tests for specific mechanical failure modes. What is the equivalent for an LLM