Geoffrey Hinton, the computer scientist widely known as the “Godfather of AI”, has called for stricter government oversight of artificial intelligence, arguing that companies should be required to demonstrate the safety of powerful AI models before releasing them to the public.
Hinton said the approach could follow the regulatory framework used for medicines in the United States, where the Food and Drug Administration (FDA) assesses products before they can be approved for public use. He also expressed concern that governments could struggle to introduce adequate safeguards quickly enough as AI technology advances.
Speaking on the Smart Girl Dumb Questions podcast hosted by Nayeema Raza, Hinton discussed the growing risks associated with AI and the role public pressure could play in forcing governments to act. Brandspur Brand News reports that he believes public awareness of the technology’s potential dangers will be crucial to securing meaningful regulatory action.
Hinton drew a comparison with climate change, arguing that political action on major risks often gains momentum when the public begins to understand their consequences. He suggested that a similar shift in public attitudes towards artificial intelligence is beginning to emerge.
His comments followed a recent closed-door briefing involving 19 United States senators. Hinton said the meeting left him somewhat more optimistic about the prospects for regulation, although he noted that most of the lawmakers present were Democrats, with 17 attending.
During the discussions, physicist Max Tegmark reportedly proposed that AI companies should face requirements similar to those imposed on pharmaceutical manufacturers, including demonstrating safety to a government regulator before releasing their products. Hinton supported the principle, describing it as a minimum standard that should be considered for artificial intelligence.
The debate comes as increasingly capable AI systems raise questions about public safety, cybersecurity and the potential misuse of advanced technology. Supporters of stronger oversight argue that relying solely on companies to assess their own products could leave significant risks insufficiently addressed.
Hinton also reflected on his own contribution to the development of AI and his decision to begin speaking publicly about its dangers. He said he did not regret conducting the research that helped advance the technology, explaining that the risks had appeared much more distant when he began his work.
His main regret, he indicated, was failing to recognise sooner how dangerous AI systems could become. He said he did not begin issuing public warnings until 2023, by which time some other researchers had already raised concerns.
The computer scientist has previously warned US lawmakers that the window for establishing effective safeguards could be limited. His latest comments reinforce his position that governments need to consider the consequences of increasingly powerful AI systems before the technology becomes more difficult to control.
Microsoft co-founder Bill Gates has also warned about potential cybersecurity and biological threats associated with AI, arguing that the risks are too serious to be left entirely to voluntary industry controls.
For businesses, governments and ordinary users, the regulatory debate could shape how AI products are developed, tested and introduced into everyday life. Stronger safety requirements could place additional responsibilities on technology companies, while potentially giving users greater confidence in systems deployed in sensitive areas.
However, questions remain over how such rules would be designed, which AI systems would require approval and how regulators could keep pace with rapid technological developments.
Hinton’s intervention adds to growing calls for governments to move beyond voluntary commitments and establish clearer safeguards. His central concern is that public pressure may prove essential in persuading policymakers to act before the risks associated with advanced artificial intelligence become harder to manage.
