DeepMind CEO proposes independent regulatory body for frontier AI models

Demis Hassabis has called for a new standards body modeled on FINRA to test and approve frontier AI models before US market release.

• Hassabis, in a post titled "A Framework for Frontier AI and the Dawning of a New Age," proposed a standards body initially relying on voluntary model review before eventual formalization.
• The proposed regulator would build on ad hoc US government reviews of Anthropic's Mythos and OpenAI's Sol, which Hassabis criticized as lacking technical expertise.
• White House AI advisor Sriram Krishnan has discounted the possibility of an AI regulator within the executive branch, stating "there will not be an FDA for AI."
• Hassabis said the regulator would be industry-funded, staffed by open-source representatives and technical experts, and could outsource evaluations to specialized AI safety groups.