Back to today’s briefing

Story perspectives

DeepMind CEO Demands US AI Watchdog by Year-End

7/15/2026

43 9 Full Breakdown

Summary

Hassabis Urges Watchdog

Demis Hassabis, Google DeepMind CEO, urged the United States to create an AI watchdog to halt dangerous frontier models.
He posted the plan on LinkedIn, outlining a watchdog that would mirror FINRA, be industry-funded, include independent open-source experts, and assess cyber, biological and deception risks.
Hassabis said he briefed the Trump administration, other labs and European officials and hopes the body will be ready before year-end.

Summary

Hassabis Warns Threats

Demis Hassabis, chief of Google AI, posted his U.S. standards-body plan on X on Tuesday.
The White House, State Department and Commerce have not responded after being asked for comment.
Hassabis and Anthropic CEO Dario Amodei urged a U.S.-led AI coalition at a G7 summit with President Donald Trump.
Hassabis warned that frontier models could soon generate nuclear and biological threats in addition to cybersecurity risks.

Summary

Krishnan Rejects FDA

White House AI advisor and a16z partner Sriram Krishnan said there will be no FDA for AI.
He suggests frontier labs share models with the Standards Body up to 30 days before release.
Funding from AI labs would let the regulator outsource evaluations to safety groups and retain open-source experts and industry talent.
Earlier U.S. reviews of Anthropic’s Mythos and OpenAI’s Sol were criticized for technical gaps and opacity.

History

Historical Context

Congress passed the Computer Fraud and Abuse Act in 1986, signed by President Ronald Reagan after the 1983 film *WarGames* raised alarm over unauthorized computer access. The law has been amended six times and narrowed by the 2021 Supreme Court decision in *Van Buren v. United States*. Demis Hassabis now cites that precedent while urging a new AI watchdog to curb emerging threats.

1 of 4

Hassabis Urges Watchdog

Demis Hassabis, Google DeepMind CEO, urged the United States to create an AI watchdog to halt dangerous frontier models.
He posted the plan on LinkedIn, outlining a watchdog that would mirror FINRA, be industry-funded, include independent open-source experts, and assess cyber, biological and deception risks.
Hassabis said he briefed the Trump administration, other labs and European officials and hopes the body will be ready before year-end.

1 / 4