Hassabis Calls for U.S. AI Standards Body to Tackle AGI Risks

Google DeepMind chief Demis Hassabis called on the U.S. to set up a federal standards body that would review frontier AI models before release. He posted the proposal on X Tuesday, arguing that “urgent action” is needed as artificial general intelligence approaches.

“We've already seen the challenges frontier models pose for cybersecurity, and other threats including nuclear and bio risks may soon emerge as capabilities continue to advance,” Hassabis wrote. The Nobel laureate heads Google's main AI research arm.

The body would operate as a public-private partnership overseen by the federal government, modeled on the Financial Industry Regulatory Authority. FINRA oversees U.S. brokerage firms and exchange markets. Hassabis said the board should include independent technical experts and open-source representatives.

Frontier labs would initially share models voluntarily for review up to 30 days before release. That would become mandatory for deployment in the U.S. market once the system proved effective, he said. Funding would come from industry and would need to be “substantial” to attract top talent and supply computing power for testing.

“Specific agentic AI tests could look for attempts to bypass safety guardrails or signs of deception, and ensure best practices, such as digitally watermarking AI-generated images and generating human-readable output tokens to understand model reasoning,” Hassabis wrote.

The White House, State Department, and Commerce Department did not immediately respond to requests for comment.

Industry leaders have been pushing for a U.S.-led coalition on AI rules. Hassabis and Anthropic CEO Dario Amodei made a similar call at a G7 meeting that included President Donald Trump last month. OpenAI's Sam Altman also argued for a standards body in a Financial Times article earlier this month.

Regulation of advanced AI models has become a flashpoint between the public and private sectors. Anthropic recently negotiated with officials after the Trump administration temporarily imposed export controls on one of its models. OpenAI faced restrictions when the U.S. government asked it to slow the rollout of a new model.

Hassabis said the U.S. is well positioned to lead because of “its economic and technical standing.”

The push comes as the AI race with China intensifies. Models from Chinese companies such as DeepSeek and Z.ai are increasingly competitive with frontier systems from Anthropic and OpenAI, according to several analysts. U.S. lawmakers are debating how to curb adoption of Chinese AI models by domestic firms. The State Department told CNBC the trend raises “serious concerns.”

For broader stock market analysis, the regulatory direction could shape which AI players benefit from clearer rules versus those that face tighter oversight.