Google employees urge CEO Pichai to reject classified Pentagon AI contract

More than 600 employees from Google’s DeepMind and Cloud units have sent a letter to Chief Executive Sundar Pichai, demanding the company refrain from entering into an agreement with the Pentagon for the use of artificial intelligence technology in classified environments.

The employees cited risks similar to those that resulted in the startup Anthropic being barred from military work earlier this year. The letter, delivered to Pichai on Monday, arrives approximately two months after Anthropic was removed from a Department of Defense list for demanding oversight mechanisms to prevent its AI from being used for domestic mass surveillance or autonomous weaponry.

In the letter, signatories argued that Google currently has no way to guarantee that its tools will not cause unmonitored harm.

“As people working on AI, we know these systems can centralize power and that they make mistakes,” the employees stated in a copy of the letter. “We feel our proximity to this technology creates a responsibility to highlight and prevent its most unethical and dangerous uses.”

A report by The Information earlier this month indicated that Google is negotiating a deal to deploy its Gemini AI models in classified environments. The proposed agreement would reportedly allow the Pentagon to utilize Google’s AI for all legal purposes.

While the parties are reportedly discussing language to prohibit the use of AI in mass surveillance or autonomous weapons operating without human control, the employees maintained that such provisions are impossible to implement in practice.

“The only way to guarantee Google is not associated with such harms is to reject any classified workloads,” the letter stated.

Google currently maintains a contract with the Pentagon via the genAI.mil platform for non-classified workloads. However, the employees warned that approving Google AI for classified work could cause “irreparable harm to Google’s reputation, its business, and its role in the world.”

One of the letter’s organizers stated in a press release that the issue largely concerns the technical protections companies can implement, noting that the Department of Defense specifically prohibits certain controls. The statement emphasized that if management is serious about preventing future harm, it must reject classified workloads for the time being.

The Pentagon and Google have yet to respond to requests for comment.

The Pentagon’s AI usage became a point of contention earlier this year after the department designated Anthropic as a “supply chain risk” because the company would not permit its models to be used for any and all legal purposes.

Following the designation—a classification typically reserved for foreign adversaries—Anthropic filed a lawsuit against the Trump administration. At the time, hundreds of employees from Google and OpenAI signed a letter supporting Anthropic.

Within hours of the late February designation against Anthropic, ChatGPT developer OpenAI signed its own agreement with the Department of Defense. This move drew rapid criticism, and CEO Sam Altman later admitted the deal “looked opportunistic and sloppy,” stating that they subsequently sought to include provisions regarding domestic surveillance.