DeepMind CEO proposes U.S.-led AI standards body for safety testing

Google DeepMind CEO Demis Hassabis is calling for the United States to establish a new standards body that would test advanced AI models for national security risks before they reach the market.

In a statement posted on X, the Nobel laureate said "urgent action" is needed to address risks associated with artificial general intelligence — AI systems that match or exceed human-level intelligence. Hassabis pointed to cybersecurity challenges already posed by frontier models and warned that nuclear and biological threats may emerge as AI capabilities advance.

The proposed framework

Hassabis envisions a federally overseen public-private partnership similar to the Financial Industry Regulatory Authority (FINRA), which regulates U.S. brokerage firms and exchanges. The body would include a board with independent technical experts and open-source representatives.

Frontier AI labs would initially share models voluntarily with the standards body for review up to 30 days before release. Once proven effective, submission would become mandatory for deployment in the U.S. market. Testing would examine whether AI systems attempt to bypass safety guardrails or show signs of deception, and would enforce best practices like watermarking AI-generated images.

The organization would require "substantial" funding to attract top technical talent and provide compute resources for large-scale testing, with industry expected to foot the bill, according to Hassabis.

Why it matters

The proposal reflects growing tension between AI companies racing to deploy powerful models and government concerns about national security. Recent export control disputes involving Anthropic and OpenAI underscore the friction between innovation speed and safety oversight. A formal standards body could provide a structured alternative to ad-hoc government interventions while maintaining U.S. competitiveness against China, where companies like DeepSeek are releasing models that rival American systems.

Industry momentum builds

Hassabis is not alone in advocating for centralized AI oversight. CNBC previously reported that he and Anthropic CEO Dario Amodei called for a U.S.-led coalition at a G7 meeting with world leaders including President Donald Trump. OpenAI CEO Sam Altman made a similar proposal in the Financial Times earlier this month.

The push for regulation comes as Chinese AI models gain traction with U.S. companies seeking to reduce costs. The State Department has told CNBC that adoption of Chinese AI systems raises "serious concerns," and lawmakers are exploring ways to limit their use by American firms.

Hassabis argued the U.S. is well-positioned to lead AI standard-setting given its economic and technical standing. The White House, State Department, and Department of Commerce have been approached for comment but have not yet responded.

These details were first reported by CNBC.