AI Won't Destroy Us, But It Will Improve Our Lives, DoE's Darío Gil ...

In the same remarks, Gil said the ratio of capability to testing at many frontier sites is "not correct." He urged labs to accelerate testing and evaluation to make models safe before release.

Gil said OpenAI, Anthropic and Google are contributing AI models and tokens to the Department of Energy's Genesis Mission. The report presents those contributions as part of the department's work with frontier AI developers. It does not specify how many models or how many tokens are involved, or the terms on which the contributions were made.

The headline framing places Gil at odds with the strongest claims that AI poses an existential threat, while still recording his concern about the pace of testing. His remarks centre on evaluation: he says the ratio of capability to testing at many frontier sites is "not correct," and that labs should accelerate testing and evaluation so that models can be made safe before release. The report links those remarks to the Genesis Mission, to which OpenAI, Anthropic and Google are contributing models and tokens.