More than 580 Alphabet employees have signed an open letter calling on Google’s chief executive to refuse classified US Department of Defense contracts involving artificial intelligence, warning that secret military deployments could prevent scrutiny of how the company’s technology is used.
More than 580 Google employees have signed an open letter urging chief executive Sundar Pichai to reject any agreement that would allow the United States Department of Defense to use the company’s artificial intelligence systems for classified military purposes.
The letter, reported by Bloomberg, was signed by staff from across Alphabet, Google’s parent company, including employees involved in artificial intelligence research. According to organisers cited in subsequent coverage by Investing.com, about two-thirds of the signatories agreed to attach their names publicly, while others chose to remain anonymous.
The protest comes amid renewed scrutiny of Google’s relationship with the Pentagon, following reports that the company has been in discussions over the possible use of its Gemini AI systems in classified defence settings. The issue has also been sharpened by a dispute between the Pentagon and Anthropic, developer of the Claude AI model, after the company sought to restrict certain military uses of its technology. The dispute was reported by The Washington Post.
“We are Google employees who are deeply concerned about ongoing negotiations between Google and the US Department of Defense,” the letter stated, according to published extracts. “As people working on AI, we know that these systems can centralise power and that they do make mistakes.”
The employees argue that classified military projects would prevent meaningful scrutiny of how Google’s AI systems are used. They said artificial intelligence should be developed for public benefit, rather than deployed in ways that could contribute to lethal autonomous weapons, mass surveillance or other harmful applications.
Their central objection is not only to military use itself, but also to the secrecy surrounding classified work. The signatories contend that, once AI models are deployed inside secure military systems, Google staff would have little or no ability to determine whether company safeguards were being observed.
“The only way to guarantee that Google does not become associated with such harms is to reject any classified workloads,” the letter reportedly said.
The dispute revives a debate that has run through Google since 2018, when thousands of employees opposed the company’s involvement in Project Maven, a Pentagon programme using AI to analyse drone footage. That internal revolt led Google to decide not to renew its contract and to adopt AI principles that included restrictions on work involving weapons or surveillance. Those principles were noted at the time by Business & Human Rights Resource Centre.
Those principles have since been revised. In February 2025, Google removed earlier language pledging not to develop AI for weapons or surveillance applications, a change reported by The Washington Post, and widely seen as giving the company greater flexibility in national security and defence work.
Google has already rebuilt parts of its defence business. Bloomberg reported in March that Google was preparing to provide AI agents to the Pentagon for unclassified work, with Gemini-based tools expected to operate across parts of the Department of Defense’s workforce.
The present letter suggests that a section of Google’s workforce sees a clear distinction between unclassified administrative or operational tools and AI systems used in secret military environments. The employees’ concern is that classified deployment would make independent oversight difficult, while exposing Google to reputational and ethical risks if the technology were used in targeting, surveillance or weapons systems.
The episode also reflects a wider shift in Silicon Valley. Major AI companies are increasingly being drawn into defence and national security contracts, as governments seek to apply generative AI to intelligence analysis, logistics, cyber operations and battlefield decision-making.
For Google, the issue is particularly sensitive because of the company’s earlier experience with Project Maven. The latest employee campaign explicitly frames the current dispute as a continuation of that earlier conflict. Organisers quoted by The Boston Globe said that “Maven is not over”, arguing that Google has still not established binding limits on the military use of its technology.
The letter does not appear to demand that Google sever all defence ties. Rather, it calls for the company to reject classified AI work where employees and outside observers would be unable to assess compliance with ethical commitments.
Google has not publicly announced a classified Pentagon AI contract. However, the scale of the employee response indicates that internal opposition to military AI work remains active inside the company, even as commercial and geopolitical pressures push the wider technology sector towards closer co-operation with defence agencies.
For Mr Pichai, the decision will test whether Google’s revised AI principles can satisfy both national security customers and a workforce that has repeatedly challenged the company’s role in military technology.
