‘AI systems cannot yet be trusted to power deadly weapons’: Anthropic rejects Pentagon demand for full military access

Leading Artificial Intelligence company Anthropic has rejected pressure from the US government to provide unrestricted access to its technology. The decision came as the US Defense Department mounted pressure on the AI firm.

Anthropic CEO Dario Amodei said they have decided to turn down the request in good conscience.

Advertisement

“These threats do not change our position: we cannot in good conscience accede to their request,” Amodei said in a statement.

It may be recalled that Anthropic was given a deadline until today (Friday, 27 February 2026) to permit unconditional military use of its technology. Washington had warned that the AI startup could be forced to follow suit under emergency federal powers if it failed to agree to the demand.

Amodei pointed out that the Pentagon and intelligence agencies have deployed Anthropic models to safeguard the country, but added that using the tools for mass surveillance of US citizens and for fully autonomous weapons would breach an ethical line.

Anthropic models cannot be allowed to be used for mass domestic surveillance as this would be incompatible with democratic values, Amodei said, adding that leading AI systems are not yet “reliable to be trusted to power deadly weapons without a human in ultimate control”.

“We will not knowingly provide a product that puts America’s warfighters and civilians at risk,” the leading AI voice stated further.

What was Pentagon’s ultimatum to Anthropic?

Pentagon officials earlier this week had held talks with Anthropic. The company was then served an ultimatum: permit unrestricted military use of its technology by 5:01 pm Friday (February 27, 2026) or face being forced to comply under the Defense Production Act.

The law, formulated during the Cold War and last invoked during the Covid-19 pandemic, gives the US federal government sweeping powers to force private industry to prioritise national security needs.

Anthropic was also warned that in case of non-compliance, it would be labelled a supply chain risk. This kind of labelling is normally given to companies from adversary countries.

Defending its demands, a senior Pentagon official was reported as saying that Anthropic’s concerns were unfounded as the Defense Department had always operated within the law. The department has only given out lawful orders, the official added.

However, Amodei’s statement further highlighted the company’s concerns: “Anthropic understands that the Department of War, not private companies, makes military decisions. However, in a narrow set of cases, we believe AI can undermine, rather than defend, democratic values.”

It may be noted that the Pentagon has approved the use of Elon Musk’s Grok system for use in a classified setting, while OpenAI and Google were also reported to be on the verge of getting similar clearances.

Last year, Anthropic, alongside other AI companies, was contracted to supply AI models for a range of military applications under a USD 200 million agreement.