Google DeepMind CEO Demis Hassabis Proposes US AI Watchdog Modeled on FINRA

According to Axios, Demis Hassabis, co-founder and CEO of Google DeepMind, is calling on the U.S. to establish a new AI watchdog with the power to screen the world's most advanced models and coordinate an industry-wide slowdown if dangers mount. In a personal manifesto published Tuesday, Hassabis proposes an AI standards body modeled on FINRA, funded by the industry, staffed by technical experts, and answerable to the U.S. government. Frontier labs would initially share models voluntarily up to 30 days before release for safety testing, with formalization to follow once the regime proves effective. The rules would apply to all frontier-class models regardless of origin. Hassabis has briefed the Trump administration, fellow lab leaders, and European officials, and says the timeline is aggressive, aiming for the new body to be operational before year-end. He warns that within 18 months, dangerous cyber, biological, and nuclear capabilities could be in open-source models beyond government control. The proposal follows the Trump administration's ad hoc crackdown on Anthropic's models last month, which Hassabis called a wake-up call.