Demis Hassabis Proposes U.S.-Led AI Watchdog for Frontier Models

Google DeepMind CEO Demis Hassabis proposed the creation of a U.S.-led AI standards body to evaluate advanced frontier AI models before they are released to the public. Outlined in a manifesto titled "A Framework for Frontier AI and the Dawning of a New Age," the proposed organization would operate as a federally overseen public-private partnership. Hassabis modeled the entity after the Financial Industry Regulatory Authority (FINRA), suggesting it be funded by the AI industry to attract technical talent and compute resources while remaining under U.S. government oversight.

The proposed body would establish safety benchmarks to mitigate cybersecurity, biological, and nuclear risks. Under this framework, AI labs would initially submit models voluntarily for review 30 days before release, with the long-term goal of making these assessments mandatory for deployment in the U.S. market. Hassabis argues that this systematic approach is necessary to replace the current case-by-case government interventions, such as the U.S. Department of Commerce's review of OpenAI's GPT-5.6 and restrictions on Anthropic's Mythos and Fable models.

While Hassabis frames the proposal as a way to ensure the safe development of artificial general intelligence (AGI), the plan has met with mixed reactions. Some critics warn that industry-funded self-regulation creates a conflict of interest or could be perceived by international partners as an instrument of U.S. strategy. Additionally, White House AI advisor Sriram Krishnan discounted the likelihood of an AI regulator being established within the executive branch.