The Pentagon threatens the AI ​​giant, they gave them a deadline of 17.01 on Friday: ‘We cannot agree to these conditions’

At the heart of the dispute is a request from the US Department of Defense that Anthropic remove security restrictions from its advanced AI model Claude, which is growing in size. used on classified military networks. The Pentagon claims freedom to use the technology for “all lawful purposes”, but Anthropic refuses to cross its two red lines: the use of AI for mass surveillance of American citizens and the development of fully autonomous weapons.

Two red lines that the Pentagon wants to erase

The conflict escalated after a tense meeting on Tuesday between Defense Secretary Pete Hegseth and Amodei, after which the company was given an ultimatum. Although the Pentagon claims it has no intention of using the technology for purposes Anthropic opposes, it insists on removing all obstacles.

​- This is a simple, common sense request that will prevent Anthropic from jeopardizing critical military operations and potentially putting our soldiers at risk. We will not allow any company to dictate the terms of how we make operational decisions – Pentagon spokesman Sean Parnell wrote on the X platform.

On the other hand, Amodei claims that today’s technology is simply not safe and reliable enough for such applications. In the press release, he stated that “some uses of AI can undermine, rather than defend, democratic values.”

“Threats do not change our position: we cannot in good conscience agree to their request,” Amodei’s statement reads.

Ultimatum and unprecedented threats

The deadline set by the Pentagon expires on Friday at 5:01 p.m. EST.. If Anthropic does not budge by then, it will face consequences that go far beyond financial loss. The first threat is the termination of the contract worth 200 million dollars, which for a company valued at around 380 billion dollars does not represent an existential threat.

Much more dangerous is that the Pentagon has threatened to declare Anthropic a “risk to the supply chain”. Such a designation is almost always used for foreign companies linked to American adversaries, and it could destroy Anthropic’s business. Any company that works with the military, such as the giants Boeing and Lockheed Martin, would have to prove that it does not use anything related to Anthropic in its work for the Pentagon, which could lead to a mass exodus of clients.

As a third measure, the Pentagon has announced the possibility of invoking the Defense Production Act (DPA), a Cold War-era measure that gives the president the power to compel private companies to cooperate for the sake of national security. Amodei pointed out the apparent contradiction in these threats.

“These two threats are fundamentally contradictory; one marks us as a security risk, and the other declares Claude to be key to national security,” he wrote.

Competitive pressure and the future of cooperation

Anthropic’s attitude is unique in the industry. Until recently, their Claude model was the only one that worked on classified military networks. However, the Pentagon has confirmed that it has reached an agreement with Elon Musk’s xAI company, and OpenAI and Google are reportedly close to similar agreements, without the ethical restrictions imposed by Anthropic.

Experts warn that this kind of pressure from the Pentagon could send a message to the rest of the industry not to try to oppose the military’s demands.

​- This is a signal to other AI companies they are negotiating with to be careful not to try to set any restrictions on the use of artificial intelligence – said Adam Connor from the Center for American Progress.

Although negotiations are at a standstill, Anthropic said it is ready to help the Pentagon with a “smooth transition to another service provider” if no deal is reached. Despiteč With a similar attitude, the company said that they still hope that Minister Hegseth will “reconsider” his decision and that they remain ready to continue the conversation.