In a post on X Tuesday morning, Google DeepMind CEO Demis Hassabis called for the creation of a new regulatory body to oversee the release of the Frontier model. The post, titled “Frontier AI Frameworks and the Dawn of a New Era,” makes the case for a “standards body” modeled after the Financial Industry Regulatory Authority (FINRA) that can test frontier models and develop best practices for their release.
“Initially, Frontier Labs voluntarily shared models with standards bodies for review up to 30 days before release,” the post reads. “Formalization could occur soon if the evaluation protocol is shown to be effective and robust, meaning the Frontier model would need to pass it before it can be deployed to the U.S. market. The Institute also plans to work with standards bodies to address critical vulnerabilities post-release.”
The proposed system builds on an ad hoc review conducted by the U.S. government of Anthropic’s Mythos and OpenAI’s Sol. These reviews generated significant criticism for a lack of technical expertise and opaque decision-making regarding when models were released. Under Hassabis’ proposed regulator, those decisions could be handed over to a new organization that would be run independently and funded by the AI industry, with support from the U.S. government.
The outlook for AI regulation remains controversial for both the tech industry and the Trump administration. Most recently, White House AI advisor and a16z general partner Sriram Krishnan dismissed the possibility of an AI regulator within the executive branch, saying, “There is no FDA for AI.”
Establishing standards bodies as self-regulatory bodies like FINRA could be a way to address these concerns. Hassabis envisions the necessary financial support from the AI lab to staff and retain industry open source representatives and technical experts at the regulator. Some assessments could also be outsourced to a growing number of AI safety groups that can specialize in specific risks.
“The strength of this approach is that it can be technologically focused while simultaneously supporting innovation and encouraging responsible behavior,” Hassabis argues. “It is designed to follow the acceleration on the ground and adapt to the greatest risks as they are identified, and can be escalated in stages as the severity of the situation requires.”
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
