Google DeepMind CEO Demis Hassabis published an X post on Tuesday morning calling for the establishment of a new independent standards body to supervise the release of frontier artificial intelligence (AI) models. The post was titled "A Framework for Frontier AI and the Dawning of a New Age."
Hassabis argues the new organisation should resemble the Financial Industry Regulatory Authority (FINRA) as a self-regulatory body that can test frontier models and develop best practices for their release. Under his proposal, Frontier Labs would initially volunteer to share models with the Standards Body for review up to 30 days before public release.
"Initially, Frontier Labs would voluntarily share models with the Standards Body for review up to 30 days before release," the post states. "Once the assessment protocol is shown to be effective and robust, formalisation could quickly follow, meaning that Frontier Models would be required to pass it to be deployed in the US market. Labs would also work with the Standards Body to address any critical post-release vulnerabilities."
The proposal builds on ad hoc reviews previously conducted by the US government of Anthropic’s Mythos and OpenAI’s Sol. Those government-led reviews faced criticism for lacking sufficient technical expertise and for opaque decision-making about when models could be released. Hassabis’s plan would transfer such decisions to a new organisation that is backed by the US government but funded by the AI industry and operated independently.
AI regulation remains contentious across the tech sector and within the Trump Administration. White House AI adviser and a16z general partner Sriram Krishnan recently dismissed the idea of an executive-branch regulator for AI, saying "there will not be an FDA for AI."
Framing the proposed body as a FINRA-style self-regulatory organisation could address some political objections. Hassabis envisages staffing the regulator with open-source representatives and technical experts from the industry, with financial support from AI labs to retain personnel. The body could also outsource some evaluations to specialised AI safety groups that focus on particular risks.
"The strength of this approach is it would be technically focused, while at the same time supporting innovation and incentivising responsible behaviour," Hassabis writes. He adds that the design would allow the organisation to keep pace with the field’s acceleration, adapt to newly identified major risks, and be scaled up if the seriousness of the situation requires it.
The post outlines the general structure and rationale for the standards body but does not provide a concrete timeline or specific legal steps for creating the organisation.
