Anthropic is opening a lawsuit against the Pentagon in an attempt to block the authority from blacklisting it over national security following the AI firm’s withdrawal from contracts with the US Department of War.

The AI firm existed its contracts with the US department following concerns about mass AI surveillance and the use of AI automation in weapons with deadly force.

Anthropic was designated as a supply chain risk by US Secretary of Defence Pete Hegseth following the startup’s refusal to remove guardrails it placed on its AI models being use for mass domestic surveillance or autonomous weapons.

Since this, however, new reports reveal resistance among workers in the Pentagon, including IT staffers. Workers are saying that they are reluctant to use alternatives to Anthropic as they consider it superior, Reuters reports.

One staffer that spoke to Reuters said that the replacement could take months, despite insistence from government to designate Anthropic a supply chain risk, and the Pentagon’s new deal with OpenAI.

Interestingly, one IT contractor which spoke to Reuters compared Claude’s AI performance to xAI’s Grok, which they said produced inconsistent responses. It remains to be seen in xAI is meant to be used by Pentagon staffers or what the prompts were, considering the new military contract with OpenAI.

The AI firm described this move as “unprecedented and unlawful” as it pushes back on the designation.

“The Constitution does not allow the government to wield its enormous power to punish a company for its protected speech”, Anthropic wrote. “No federal statute authorizes the actions taken here.”

The lawsuit will target the US executive office, which includes President Donald Trump and his cabinet, including Hegseth and Secretary of State Marco Rubio, and 16 government agencies.

Anthropic alleges that its negotiations on the terms and conditions with the US Department of War, otherwise known as the Department of Defence, was cut off after it tried to limit its AI use for surveillance and autonomous weapons.

The designation persists and the lawsuit is set to proceed even though Anthropic was still being employed by the Pentagon in its bombing of Iran.

Following the designation, OpenAI swooped in to be contracted by the Pentagon, with CEO Sam Altman alleging that its deal with the US government had better assurances against surveillance and autonomous weapons.

Recommended reading

• OpenAI Backtracks on “Opportunistic and Sloppy” US Government Deal
• Anthropic Out, OpenAI In | The US AI Power Struggles
• Anthropic Reaches $380bn as AI Investment Frenzy Rolls on

However, this has continually been challenged, especially now as OpenAI’s robotics head Caitlin Kalinowski resigned from the company over the deal.

She described the deal as “rush” in her public resignation post, saying that the contract did not have robust guardrails against surveillance.

“We think our agreement has more guardrails than any previous agreement for classified AI deployments, including Anthropic’s,” OpenAI wrote in a statement.

“In our agreement, we protect our red lines through a more expansive, multi-layered approach. We retain full discretion over our safety stack, we deploy via cloud, cleared OpenAI personnel are in the loop, and we have strong contractual protections.”