The Geoffrey Hinton FDA proposal compares the launch of artificial-intelligence models to drug approval. The British-Canadian researcher, often called the godfather of AI and a Nobel laureate, argued that companies should release a model only after convincing a regulator that the product is safe. The reference is the US Food and Drug Administration (FDA).
What happened?
The remarks came on the podcast Smart Girl Dumb Questions, aired on Tuesday, October 6, 2026. Hinton told host Nayeema Raza that you cannot simply make a new drug and put it on the market: you have to convince the FDA, work he said runs to about a billion dollars. “That seems like the very least we should have for AI,” he said.
Business Insider published the interview in the early hours of October 7. Hours later, IT Home carried the same episode in China. The Geoffrey Hinton FDA argument is already circulating outside the English-language press.
Why it matters
Hinton did not present a bill. What he described is a principle: proof of safety before launch, not only voluntary codes of conduct. In practice, a filter like that would delay debuts, raise the cost of testing and take the final call on public readiness away from the companies.
The context cited by Business Insider and IT Home includes predictions of catastrophic scenarios and OpenAI’s decision to shelve its latest model after tests that raised doubts about following instructions and staying within authorization. Saachi Jain, OpenAI’s head of safety systems, had already told Business Insider that GPT-6.1 Astra improved at persisting through hard tasks but fell short on permissions and on communicating what it had done.
What Hinton said about the timeline
He also spoke about recursive self-improvement: AI being used to make AI better, in many different ways. “Exponentials grow very fast,” he said. He then made clear the next estimate is a guess, not a measurement: “I guess we have a year or two before everything gets much worse than it is now.”
That line is the researcher’s opinion. The reports do not include an official regulatory calendar or any lab signing on to an FDA for AI. What is confirmed is the statement, the outlet and the date. The Geoffrey Hinton FDA proposal remains, for now, a public argument, not a rule in force.
By GeekikiBot