Google and the Pentagon sign classified deal to give the Department of Defense unfettered access to its AI models

Source: Engadget

Deal Overview

Google has signed a classified agreement that allows the U.S. Department of Defense to use its AI models for “any lawful government purpose,” according to a report by The Information. The full details of the contract remain classified.

Conditions and Oversight

• An anonymous internal source says the agreement includes a clause that Google’s AI technology should not be used for domestic mass surveillance or autonomous weapons without appropriate human oversight and control.
• The contract reportedly does not give Google any right to control or veto the government’s use of the technology.
• A Google spokesperson told Reuters that providing API access to commercial models “with industry‑standard practices and terms” is a “responsible approach to supporting national security,” and reiterated the company’s stance that AI should not be used for mass surveillance or autonomous weaponry without human oversight.

Employee Opposition

Nearly 600 Google employees signed an open letter to CEO Sundar Pichai, urging the company to reject the deal. The letter argues that the technology could be used in “inhumane or extremely harmful ways” and cites concerns about civil liberties and the potential for AI systems to centralize power and make mistakes.

“Human lives are already being lost and civil liberties put at risk at home and abroad from misuses of the technology we are playing a key role in building,” the letter states.

Industry Context

Google joins other AI firms that have entered classified agreements with the U.S. government:

• OpenAI – see the deal details in this Engadget article.
• xAI (Elon Musk’s company) – see coverage in this Engadget story.

Anthropic previously had a deal but refused to remove safeguards related to weapons and surveillance. This refusal led to a backlash from the Pentagon and President Trump, resulting in Anthropic being blacklisted from federal use.

Government Response

The agreement’s language about “appropriate human oversight and control” appears at odds with the Pentagon’s actions toward Anthropic, raising questions about the consistency of oversight policies across different AI providers.

Engadget has reached out to Google for additional details and will update this post when more information becomes available.