Demis Hassabis pitched a new independent industry AI safety entity to top Trump officials before stepping down as DeepMind CEO, according to a Wall Street Journal report summarized by Techmeme. The proposed entity was modeled on the International Atomic Energy Agency, the report says. Techmeme’s summary also says Hassabis discussed the idea with heads of other AI labs. The available summary does not provide the proposed body’s mandate, funding model, membership, enforcement powers, or whether Trump administration officials supported the idea. It also does not name the other AI lab leaders involved in the discussions. That makes this a developing governance story rather than a settled policy move. What is clear from the report is that a leading AI executive was exploring an independent safety institution outside any single company, and was raising it with both government officials and other major labs. Who benefits: Major AI labs could benefit if an industry safety entity creates shared rules or reduces uncertainty around government oversight. The current evidence does not show whether smaller labs, open-source developers, or cloud providers would be included. Who's exposed: AI companies would be exposed if a future safety body gained real auditing, reporting, or compliance authority. For now, the report only establishes that the idea was pitched and discussed, not adopted.