Anthropic Says It Rejected Pentagon’s AI Demands

Anthropic on Feb. 26 rejected the Pentagon’s request to allow unrestricted use of its Claude AI model, citing concerns that the technology could be used for mass domestic surveillance or fully autonomous weapons.

Anthropic CEO Dario Amodei said in a lengthy blog post that use cases such as mass surveillance and the development of autonomous weapons have never been included in the company’s contracts with the Pentagon.

Amodei said those two applications are “simply outside the bounds of what today’s technology can safely and reliably do,” and warned that, in a narrow set of cases, artificial intelligence (AI) may erode democratic values.

“It is the Department’s prerogative to select contractors most aligned with their vision. But given the substantial value that Anthropic’s technology provides to our armed forces, we hope they reconsider,” he said.

According to Amodei, the Department of War has threatened to remove Anthropic from its systems if the company refuses to remove safeguards related to the two uses in question.

Amodei said the Pentagon also threatened to designate Anthropic as a “supply chain risk,” which he said in his blog post is generally applied to U.S. adversaries.

He also alleged that the Department of War threatened to invoke the Defense Production Act to compel Anthropic to allow unrestricted use of its technology if the company refused.

“These latter two threats are inherently contradictory: one labels us a security risk; the other labels Claude as essential to national security,” Amodei wrote, adding that Anthropic will not yield to the pressure.

“Regardless, these threats do not change our position: we cannot in good conscience accede to their request.”

Amodei said the company hopes to continue supporting the military with the requested safeguards in place.

He also said that Anthropic is prepared to make a smooth transition to another provider if the Pentagon decides to end its partnership.

The Pentagon uses the Claude AI system for mission-critical functions, including intelligence analysis, modeling and simulation, operational planning, and cyber operations, according to Anthropic.

In a social media post, Pentagon spokesperson Sean Parnell said that mass domestic surveillance is illegal and that the Department of War has no intention to use AI for such purposes.

Parnell said the Pentagon is not seeking to use AI to develop autonomous weapons that can function without human involvement. He clarified that the military only asked Anthropic to allow it to use Claude models for “all lawful purposes.”

“This is a simple, common-sense request that will prevent Anthropic from jeopardizing critical military operations and potentially putting our warfighters at risk,” he said in a post on X.

Parnell said Anthropic is given until 5.01 p.m. ET on Feb. 27 to make a decision, warning that failure to come to an agreement would lead the Pentagon to terminate its partnership and deem the company “a supply chain risk.”

“We will not let ANY company dictate the terms regarding how we make operational decisions,” Parnell wrote.

The Pentagon last July awarded Anthropic a $200 million contract that requires the company to prototype frontier AI capabilities to advance U.S. national security.