Wire · regulatory
Exclusive: Google DeepMind's Demis Hassabis calls for U.S.-led global AI watchdog
◆ Sectors
◆ Geography
◆ Source
◆ Verified
Fusion42 · 15 July 2026 · Fusion42 review
Google DeepMind CEO Demis Hassabis is calling for a U.S.-led global AI watchdog modelled on FINRA, with power to screen frontier models before release and coordinate industry-wide slowdowns if risks mount. The proposal—backed by Trump administration officials and other major lab leaders—would require voluntary model sharing initially, moving to mandatory pre-deployment testing within months, with a board of independent experts, government and industry representatives.
This Wire brief sits within Fusion42's coverage of AI Frontier Models.
◆ ◆ The Wire takeaway
If you're building frontier AI models, your release calendar just got a mandatory 30-day checkpoint before you can ship in the U.S. market—and the rules will tighten as capabilities evolve. The ad hoc crackdowns on Anthropic and OpenAI are being replaced by a formal regime, which means the uncertainty is ending, but so is your freedom to deploy without proof.
◆ Coverage
1 source · 15 Jul 2026
◆ Related on Wire
◆ Topics