Demis Hassabis, co-founder and CEO of Google DeepMind, has called for the establishment of a U.S.-led regulatory body to oversee advanced artificial intelligence (AI) models. In a manifesto published on July 14, 2026, he emphasized the need for a systematic approach to AI regulation, which would be funded by the industry and staffed by technical experts accountable to the U.S. government.
Hassabis highlighted that current AI-driven cyber risks serve as warning signs, suggesting that more severe threats could emerge from advanced models within 18 months. He noted that the proposed regulatory body would initially allow frontier labs to voluntarily share their models for safety testing before public release. The goal is to ensure that AI models meet safety standards before being deployed in the U.S. market.
Hassabis has been building support for this initiative over several months, engaging with officials from the Trump administration and other industry leaders. He believes that the establishment of this regulatory body is crucial for addressing the evolving challenges posed by AI technologies. The proposed framework would apply to all frontier-class models, regardless of their origin, and would involve regular updates to the qualifying benchmarks as capabilities advance.