Google DeepMind scientist backs Anthropic limits in Pentagon dispute

In short: Google DeepMind chief scientist Jeff Dean publicly supported Anthropic’s refusal to remove limits on how the Pentagon can use its Claude AI model.

What happened

Jeff Dean, the chief scientist at Google DeepMind, posted on X on February 25, 2026 in support of Anthropic’s stance in a dispute with the US Department of Defense. Anthropic makes Claude, a chatbot style AI system, and has said it will not allow the model to be used for fully autonomous weapons or for mass surveillance of Americans.

The dispute centers on contract language. According to Wired, Defense Secretary Pete Hegseth gave Anthropic an ultimatum to remove safeguards by Friday evening or face contract cancellation, a “supply chain risk” label, and possible use of the Defense Production Act (a law that can force companies to prioritize government needs).

Dean echoed concerns raised by Harvard professor Boaz Barak, including that using AI for mass surveillance could violate the Fourth Amendment, which protects Americans against unreasonable searches. He also said his position has not changed since he signed a 2018 pledge opposing lethal autonomous weapons, writing that the decision to take a human life should not be handed to a machine.

Why it matters

This is an unusual moment where a top AI leader at one major lab is publicly backing another lab against military pressure. If the government gets “any lawful purpose” access, it could set a template, like a standard contract clause, that other AI companies may be pushed to accept. The disagreement also shows growing internal pressure, including employees urging limits for Google’s own AI model, Gemini.

Source: Wired