Demis Hassabis, CEO of Google DeepMind, is calling for the creation of an international AI oversight body with real teeth — the power to slow or stop the release of frontier models if they pose serious risks. In a blog post published this week, the Nobel laureate argued that the United States should lead this effort, citing the country’s economic and technical clout as the natural foundation for setting global norms.

The proposed organization, modeled on regulators like the Financial Industry Regulatory Authority, would consist of independent experts and open-source community representatives. Its mandate: evaluate advanced AI models before deployment and, if necessary, coordinate an industry-wide pause. Hassabis warns that artificial general intelligence may arrive within a few years, describing the moment as "the foothills of the singularity" and a new chapter for humanity. He has been quietly rallying support from the Trump administration, rival AI labs, and European officials, and told Axios he aims to have the body operational before 2025.

The push comes amid a regulatory vacuum: no global or comprehensive U.S. framework currently governs AI. Hassabis, who last month signed a statement urging stronger safeguards against AI-enabled bioweapons, joins a growing chorus of tech leaders — including Anthropic’s Jack Clark and former Google CEO Eric Schmidt — who warn that the economic and security implications of advanced AI demand urgent, coordinated action.

Source: The Verge