Google broadens Pentagon access to its AI following Anthropic’s refusal

Google has expanded access to its artificial intelligence systems for the U.S. Department of Defence, enabling use within classified networks and permitting a wide range of lawful applications, according to multiple reports.

This development comes after Anthropic declined to provide similar access under the same conditions. The Pentagon had sought fewer restrictions on how AI systems could be used. At the same time, Anthropic pushed for safeguards to prevent applications such as domestic mass surveillance and the deployment of autonomous weapons.

Following Anthropic’s refusal, the Department of Defence labelled the company a “supply-chain risk,” a classification typically associated with foreign adversaries. The dispute has since escalated into legal action, with Anthropic challenging the designation. A judge recently granted the company an injunction, allowing the case to proceed while temporarily blocking the classification.

Google’s move makes it the third major AI provider to step in after Anthropic’s decision. OpenAI quickly agreed with the Department of Defence, and xAI followed with a similar deal.

According to reporting by The Wall Street Journal, Google’s agreement includes language indicating that the company does not intend for its AI systems to be used for domestic mass surveillance or autonomous weapons. Comparable provisions are said to exist in OpenAI’s agreement as well. However, questions remain about whether these conditions are legally binding or enforceable.

The agreement has also drawn internal criticism. Nearly 950 Google employees have signed an open letter urging the company to adopt a position similar to Anthropic’s and avoid providing AI technology to the Defence Department without stricter safeguards. Google has not publicly responded to requests for comment regarding the concerns.