Anthropic clashes with Trump administration after Pentagon brands it a supply-chain risk, as AI firm refuses to allow mass surveillance and autonomous weapons use.
A sharp confrontation has erupted between Anthropic and the Trump administration after the Pentagon moved to designate the artificial intelligence firm a “supply-chain risk”, prompting the company to denounce the decision as “legally unsound” and unprecedented.
The dispute centres on demands from the Department of War, led by Secretary Pete Hegseth, for unrestricted access to Anthropic’s flagship AI model, Claude. The company, founded by Dario Amodei, says negotiations broke down over two specific carve-outs: mass domestic surveillance of Americans and the deployment of fully autonomous weapons.
In a strongly worded statement, Anthropic said it had engaged in months of good-faith discussions with the department but could not agree to remove safeguards preventing those uses. “We believe that mass domestic surveillance of Americans constitutes a violation of fundamental rights,” the company said, adding that fully autonomous weapons posed serious ethical and reliability concerns.
The administration escalated matters after President Donald Trump ordered federal agencies to cease using Anthropic’s technology. Hegseth subsequently directed the Department of War to label the firm a supply-chain risk to national security, declaring that no contractor or partner working with the US military could continue commercial activity with Anthropic. He framed the move as essential to ensuring “full, unrestricted access” to AI systems for all lawful defence purposes.
Anthropic countered that such a designation would represent a historic departure from established practice. Supply-chain risk determinations have typically been reserved for foreign adversaries, not American companies engaged in contractual negotiations with the government. The company further argued that the defence secretary lacks statutory authority to impose the sweeping commercial restrictions suggested.
Citing 10 USC 3252, Anthropic said any supply-chain risk designation would legally apply only to the use of its Claude model within Department of War contracts and could not extend to contractors’ broader commercial relationships. “No amount of intimidation or punishment will change our position on mass domestic surveillance or fully autonomous weapons,” the company added.
Hegseth, posting on social media, accused Anthropic of “duplicity” and ideological grandstanding, claiming the firm had attempted to strong-arm the US military by refusing certain applications. He insisted the department’s stance would not waver and announced a six-month transition period for defence-related services to shift to alternative providers.
