DEEPMIND’S DEMIS HASSABIS DEMANDS A FINRA FOR AI — WANTS EVERY FRONTIER MODEL TESTED BY INDEPENDENT EXPERTS BEFORE IT CAN SHIP Google DeepMind CEO Demis Hassabis went public Monday with a call for the United States to create an independent regulatory body modeled after FINRA, the Financial Industry Regulatory Authority, specifically designed to test and certify frontier AI models before they can be deployed to the public. He told Axios he wants the body operational before the end of this year. The proposal is more detailed than the usual Silicon Valley platitudes about safety. Hassabis envisions a body funded by industry but staffed by independent technical experts, Turing Award winners and top researchers, alongside government and open-source representatives, with a majority-independent board insulated from industry capture. Frontier labs would submit models up to 30 days before release for probing of dangerous capabilities across cyber, biological, and deception categories. Critically, Hassabis said the rules should apply to all frontier models regardless of country of origin. That means Chinese models would have to pass the same bar or face exclusion from the US market. Once voluntary participation proves effective, he envisions formal legal requirements following quickly. This is the CEO of one of the world’s most powerful AI labs publicly telling regulators to regulate his own industry. That sentence alone tells you how close to the edge they are all operating. Hassabis has seen the capabilities coming out of DeepMind’s labs. The urgency in his proposal reflects what he knows is already in the pipeline. Keywords: DeepMind AI regulation, Demis Hassabis AI watchdog, frontier AI standards body, AI regulation 2026