Demis Hassabis on Tuesday published a manifesto calling on the U.S. to establish a new industry-funded AI watchdog to screen the world’s most advanced models and coordinate industry slowdowns if dangers mount.
Hassabis, Google DeepMind co-founder and CEO and the Nobel laureate behind Gemini, published the manifesto titled "A Framework for Frontier AI and the Dawning of a New Age."
He proposed modeling the body on FINRA (Financial Industry Regulatory Authority) and said frontier labs would initially share their models voluntarily up to 30 days before release for safety testing that probes dangerous cyber, biological and "deception" capabilities.
Hassabis called today's AI-driven cyber risks "warning shots" and warned that within 18 months those capabilities — plus far graver biological and nuclear threats — could live inside open-source models beyond any government's control; he said the new body should be operational "before year-end."
He said he spent months briefing the Trump administration, fellow lab leaders and European officials about the plan and that "The noises I've been hearing are very positive."
Hassabis cited the Trump administration's improvised crackdown on Anthropic's Mythos and Fable models last month, saying those models were frozen overnight by an export-control order and that Anthropic then spent 2½ weeks negotiating their release; he noted OpenAI agreed to restrict GPT-5.6 to government-vetted partners at launch and that it was released publicly last week after negotiations and testing with the Commerce Department.
Anthropic CEO Dario Amodei has called for binding regulation envisioning an FAA-style agency with the power to block unsafe models, and Hassabis said the lab chiefs behind Gemini and Claude now agree Washington should regulate them while differing mainly on who holds the authority.
Hassabis wrote that AGI is "probably only a few short years away" and that "We've essentially found a way to make sand think," adding, "It's miraculous."