US military wants unrestricted access to cloud AI, but Anthropic is refusing
Anthropic, the company behind the AI Model Cloud, has been in controversy over US government restrictions on the use of AI. Now, according to reports, the Dario Amodei-led startup is unlikely to accept US Defense Secretary Pete Hegseth’s Friday deadline.
Anthropic, one of the world’s leading AI companies, is facing a deadline. The Dario Amodei-led startup has until Friday for the Pentagon to accede to US Defense Secretary Pete Hegseth’s demands to lift the ban on its AI models. However, a new report indicates that Amodei is unwilling to bow down.
According to Reuters, Anthropic has “no intention” of relaxing its use restrictions when it comes to military purposes. This comes after Pete Hegseth called in Dario Amodei to discuss removing these security measures.
❮❯
Why is Anthropic at odds with the US government?
The controversy centers on whether Anthropic should allow its systems to be used for unrestricted military purposes, including fully autonomous weapons and mass surveillance of Americans. According to the report, Dario Amodei described these as “red lines”, reiterating the company’s commitment to security-focused AI development.
Earlier, the US government had said that it had used Anthropic’s AI model during its operation to capture Venezuelan President Nicolas Maduro.
What will the US government do if Anthropic fails to meet the deadline?
During the meeting with Amodei, Pete Hegseth warned that Anthropic could face significant consequences if it did not comply. According to reports, Hegseth said the Pentagon would either declare Anthropic a “supply chain risk” – a designation usually reserved for foreign adversaries – or invoke the Defense Production Act (DPA) to force the company to tailor its AI models to military needs.
If Anthropic is labeled as a supply chain risk, all US government defense contractors will be required to sever ties with Anthropic. The Amodei-led startup signed a $200 million contract with the US government last year.
On the other hand, DPA enables the US government to force companies to prioritize or expand production for national defense.
Pentagon officials argue that military use of the technology should be governed by US law and constitutional limits, not private company policies. A senior Pentagon official told The Associated Press that “the department issued only lawful orders and any legal responsibility for the use of the tools rests with the Army, not the AI provider.”
Anthropic is the only major provider approved for classified military networks, which could make things difficult for the US Department of Defense if they are forced to cut ties with the AI startup. However the Pentagon has started signing contracts with other AI firms like xAI.
