This move could lead to the cancellation of a contract worth up to 200 million USD between the two sides.
Controversy broke out after Anthropic refused to remove technical barriers that prevented the company's technology from being used to control automatic weapons or deploy surveillance in the US territory.
Pentagon spokesman Sean Parnell wrote on social network X that the Department of War has no intention of using AI to monitor a series of American citizens, nor does it want to develop automatic weapons that operate completely without human control.
He said the Pentagon's request to use Anthropic's AI model for all legitimate purposes and set a deadline for the company to make a decision. If it does not agree, the Pentagon will terminate cooperation and consider Anthropic a "supply chain risk".
In a statement on February 26, Anthropic CEO Dario Amodei affirmed that the company opposes the use of its model for large-scale domestic surveillance or fully automated weapons operation. He said that advanced AI systems today are not reliable enough for vital tasks.
A source close to the company said that Anthropic did not accuse the Pentagon of planning to deploy AI in these directions, but only gave an assessment of technology safety.
According to sources, AI can behave unpredictably in new situations, leading to the risk of misfiring, mission failure, or unintended escalation. The use of AI for large-scale surveillance is also worrying because current law does not clearly limit the inferences that AI can draw from big data.
Mr. Amodei said that if the contract is cancelled, Anthropic will support the transfer process to another supplier. However, he emphasized that pressure from the Pentagon, including threats to remove the company from the system or citing the Defense Production Act, will not change the company's position.
Deputy Secretary of War Emil Michael responded on X, criticizing Mr. Amodei and affirming that the Pentagon always complies with the law, and does not allow any profit-seeking technology companies to dominate defense policy.
Anthropic, backed by Google and Amazon, currently has a contract with the US Department of War worth a maximum of 200 million USD. The company said it is still ready to continue dialogue and ensure operations are not interrupted.
More than 200 employees of Google and OpenAI have signed open letters supporting Anthropic's position. Google and OpenAI have not made immediate comments.
Anthropic is an artificial intelligence company based in San Francisco, founded in 2021 by former OpenAI leaders, including Dario Amodei. The company develops AI models like Claude and receives large investments from Google and Amazon.
