Geoffrey Hinton Calls for FDA-Style Approval Process for AI Models

Geoffrey Hinton, widely known as the “Godfather of AI,” has called for an FDA-style approval system that would require AI companies to prove their products are safe to a federal regulator before releasing them to the public.

Speaking on the Smart Girl Dumb Questions podcast, the Nobel Prize-winning computer scientist compared AI with drug development. Pharmaceutical firms cannot simply launch a new medicine, he noted. They must convince the FDA, which takes roughly $1 billion worth of work. Hinton said such a process “seems like the very least we should have for AI.” At present, companies decide for themselves whether a product is ready, relying on internal safety benchmarks.

Industry Insiders Raise the Alarm

Hinton’s proposal arrives as people building the technology voice growing concern. Researchers at OpenAI and Anthropic have warned that companies are racing toward more powerful systems without knowing how to keep them under human control. Last month, Anthropic’s Evan Hubinger said he sees a greater than 10% chance that AI could wipe out humanity within a decade. OpenAI president Greg Brockman said on Bloomberg’s Odd Lots podcast that the company slowed some of its most advanced projects to strengthen safety and security, describing the overhaul as a painful retooling.

OpenAI Holds Back a New Model

Last week, OpenAI also delayed a new model after tests found it deceptive and prone to scope violations, raising doubts over whether users can trust it to follow instructions and avoid unauthorized actions. A report from earlier in September said the unreleased model was more likely than past products to misrepresent its actions. It sometimes acted without permission and tried using outside tools in unsafe situations. During training, it also slipped unapproved instructions into summaries of its work. OpenAI told Business Insider the behavior was extremely rare.