Anthropic will not budge as the Pentagon ramps up the AI ​​conflict

Anthropic has until Friday evening to give the US military unlimited access to its AI model or face the consequences, reports Axios.

Secretary of Defense Pete Hegseth told Anthropic CEO Dario Amodei in a meet on Tuesday morning that the Pentagon will declare that Anthropic is a “threat of support products” – a designation usually reserved for foreign enemies – or request the Defense Production Act (DPA) to force the company to adapt to the needs of the military.

The DPA gives the president the power to compel companies to prioritize or expand national defense production. It was recently called upon during the COVID-19 pandemic to force companies like General Motors and 3M to produce ventilators and masks, respectively.

Anthropic has long said that it does not want its technology to be used for surveillance of Americans or autonomous weapons — and it refuses to budge on these facts.

Pentagon officials say the military’s use of the technology should be governed by US law and constitutional limits, not by private contractor spending policies.

The use of the DPA in a dispute over the protection of AI would represent a significant growth in the use of modern law. It will also reflect a growing sense of agency instability that has grown exponentially in recent years, according to Dean Ball, a senior fellow at the Foundation for American Innovation and a former adviser on AI policy in the Trump White House.

“It would be the government saying, ‘If you don’t agree with us politically, we’re going to try to fire you,'” Ball said.

The debate is taking place against an ideological divide, with some in the administration – including AI chief David Sacks – openly criticizing it. Anthropic security policies such as “woke”.

“Any reasonable, responsible businessman or corporate manager will look at this and think that the US is not a sustainable place to do business,” Ball said. “This is in stark contrast to what makes America the business center of the world. We have always had stable and reliable laws.”

It’s a dangerous game of chicken, and Anthropic might not be the first to blink. According to ReutersAnthropic does not plan to reduce its usage restrictions.

Anthropic is the DOD’s only limited AI lab, according to multiple reports. The Department of Defense doesn’t have a backup plan in play — even the Pentagon does he says contracted to use xAI’s Grok in selected systems.

The lack of redundancy may help explain the Pentagon’s aggressive behavior, Ball argued.

“If Anthropic cancels the contract tomorrow, it would be a huge blow to the DOD,” he told TechCrunch, noting that the agency appears to be failing. National Security Memorandum from the end of the Biden administration that directs federal agencies to avoid relying on a single AI border planning system.

“DOD has no backup. This is a one-time transaction,” he continued. “They can’t fix this overnight.”

TechCrunch has reached out to Anthropic and DOD for comment.