Anthropic refuses Pentagon AI access as a high-stakes standoff between the AI company and the U.S. Department of War approaches a critical deadline.

With the Pentagon pushing for unrestricted use of Anthropic’s technology, hundreds of employees from rival firms Google and OpenAI have signed an open letter urging their leaders to stand with Anthropic and reject the military’s demands.

Anthropic Refuses Pentagon AI Access Over Surveillance and Weapons

At the center of the dispute is Anthropic’s refusal to allow its systems to be used for AI mass surveillance or fully autonomous weapons. The company has drawn a clear boundary: its models should not be deployed for domestic surveillance of Americans or weapon systems without a human in the loop.

The open letter, signed by more than 300 Google employees and over 60 OpenAI employees, calls on their executives to “put aside their differences and stand together” in support of those red lines.

“They’re trying to divide each company with fear that the other will give in,” the letter states, warning that fragmentation among AI firms could weaken collective resistance.

As of publication, executives at Google and OpenAI have not formally responded to the letter. However, informal statements suggest alignment with Anthropic’s stance.

In an interview with CNBC, OpenAI CEO Sam Altman said he does not believe the Pentagon should threaten companies with the Defense Production Act AI mechanism to compel compliance. According to CNN reporting, an OpenAI spokesperson confirmed that the company shares opposition to autonomous weapons and AI mass surveillance.

Google DeepMind has not issued a corporate statement, but Chief Scientist Jeff Dean expressed personal opposition to government surveillance, writing on X that mass surveillance violates constitutional protections and can chill free expression.

Defense Production Act AI Threat Escalates Tensions

Defense Secretary Pete Hegseth reportedly warned Anthropic CEO Dario Amodei that failure to comply could result in the company being labeled a “supply chain risk” or facing invocation of the Defense Production Act AI authority.

The Defense Production Act allows the U.S. president to require companies to prioritize or expand production for national defense purposes.

Amodei responded by pointing out what he described as a contradiction: one threat frames Anthropic as a security risk, while the other deems its technology essential to national security.

In a public statement, he reaffirmed the company’s position, saying it cannot in good conscience grant the Pentagon unrestricted access.

Broader Industry and Policy Implications

Anthropic already has an existing partnership with the Pentagon, but the current dispute centers on scope and safeguards. According to Axios, the military can use systems such as X’s Grok, Google’s Gemini, and OpenAI’s ChatGPT for unclassified tasks, and negotiations are underway for classified use with some providers.

Policy researchers have long debated the implications of AI mass surveillance. Studies from institutions like the Electronic Frontier Foundation and academic legal scholars warn that large-scale surveillance systems can introduce civil liberties risks and discriminatory outcomes if not tightly controlled.

The standoff highlights a broader question for the AI industry: how much influence should governments have over frontier AI systems developed by private companies?

As the deadline looms, Anthropic refuses Pentagon AI access under current terms, drawing a line that may define the next phase of AI governance in national security contexts.