Google DeepMind CEO Demis Hassabis called on Tuesday for the creation of an independent standards body to test and regulate frontier AI models, modeled after the Financial Industry Regulatory Authority (FINRA) that oversees US securities firms. The proposal, laid out in a post titled A Framework for Frontier AI and the Dawning of a New Age, calls for a US-led organization that is government-backed but industry-funded and independently operated.
Under the plan, frontier labs would voluntarily submit models for review up to 30 days before release, with staff drawn from open-source representatives and technical experts evaluating risks in areas like cybersecurity and biology research. Once the assessment protocol is proven effective, compliance could become mandatory for deployment in the US market. Hassabis said he hopes to launch the body by years end.
The proposal responds to criticism of recent ad hoc government reviews of models from Anthropic and OpenAI, which drew complaints about a lack of technical expertise and opaque decision-making. It also threads a political needle, avoiding the FDA for AI framing the Trump administration has opposed. OpenAI CEO Sam Altman publicly backed the idea, calling it a thoughtful proposal.