In a post on X on July 14, Google DeepMind CEO Demis Hassabis called for the creation of an independent standards body to regulate the release of frontier AI models. The proposed framework, titled 'A Framework for Frontier AI and the Dawning of a New Age,' draws inspiration from the Financial Industry Regulatory Authority (FINRA) and aims to test models and develop best practices for their deployment. Hassabis outlined a voluntary process where Frontier Labs would share models with the Standards Body for review up to 30 days before release. Once the assessment protocol proves effective, formal requirements could be enforced for models to be deployed in the U.S. market.
The initiative would build on previous ad hoc reviews of models like Anthropic’s Mythos and OpenAI’s Sol, which faced criticism for lacking technical expertise and transparency in decision-making. Hassabis envisions a regulator backed by the U.S. government but funded by the AI industry, with technical experts and open-source representatives staffing it. This approach would allow for specialized evaluations by AI safety groups, ensuring the system keeps pace with the field’s rapid development and adapts to emerging risks.
Hassabis argues that the standards body would focus on technical aspects while supporting innovation and responsible behavior. He emphasized that the system could be adjusted based on the severity of risks. The proposal reflects ongoing debates over AI regulation, with the Trump administration and tech industry remaining skeptical of centralized oversight. Hassabis’ plan aims to address these concerns by establishing a self-regulatory model similar to FINRA.
Source: techcrunch
