Pentagon Cuts Ties with AI Company Anthropic Over Military Use Concerns

Defense Secretary Pete Hegseth has ended the Pentagon's partnership with AI company Anthropic after the firm refused to allow its technology to be used for mass surveillance or autonomous weapons. The unprecedented move designates the U.S. company as a supply chain risk, prompting Anthropic to vow legal action while competitor OpenAI steps in to fill the void.

A major confrontation between the Pentagon and artificial intelligence company Anthropic has become public this week, with Defense Secretary Pete Hegseth abruptly cutting off the military’s partnership with the tech firm and other federal agencies.

The conflict began when Anthropic CEO Dario Amodei stood firm on his company’s position against allowing their AI products to be utilized for widespread surveillance operations or fully automated military drones. This stance led President Donald Trump and Hegseth to claim the San Francisco-based AI company poses a threat to America’s national security.

In an unusual step, Hegseth moved to classify Anthropic as a supply chain threat – a designation typically reserved for foreign companies with ties to hostile nations. The company has promised to fight back in court against what they describe as an legally questionable action.

“We will challenge any supply chain risk designation in court,” Anthropic stated, calling it an action “never before publicly applied to an American company.”

This brewing court fight could significantly impact the technology industry’s power structure and establish new precedents for how the military can use AI technology, along with safety measures designed to protect human lives.

The dispute has already created a major win for OpenAI, the company behind ChatGPT, which quickly moved to offer its services to the Pentagon after Anthropic rejected the Trump administration’s conditions. This development is expected to intensify the rivalry between OpenAI’s Sam Altman, who faced his own board troubles in 2023, and Amodei, who departed OpenAI in 2021 to start Anthropic due to AI safety concerns.

The Pentagon’s decision to label Anthropic as a defense supply chain threat will terminate a contract worth up to $200 million with the AI firm. Military officials say this action will also prevent other defense contractors from partnering with Anthropic.

Trump announced on Truth Social that most federal agencies must immediately cease using Anthropic’s AI systems, though he granted the Pentagon six months to remove the technology already integrated into military systems.

Anthropic disputes Hegseth’s legal power to block business relationships with other defense contractors. The company maintains that firms with existing commercial agreements can continue using Anthropic’s products for non-military projects.

The supply chain risk classification was originally established to help military leaders reduce the Pentagon’s dealings with potentially dangerous companies. This list has historically featured businesses connected to rival nations, including Chinese telecom company Huawei and Russian cybersecurity firm Kaspersky.

For Anthropic, this designation sends a clear message to other AI and defense companies about the consequences of refusing government demands.

“We don’t need it, we don’t want it, and will not do business with them again!” Trump declared on social media.

The six-month transition period Trump established essentially provides time for other companies to obtain the necessary classified security approvals to work with the Pentagon.

Anthropic reports they have not yet received official notification of Hegseth’s designation.

“When we receive some kind of formal action, we will look at it, we will understand it and we will challenge it in court,” Amodei promised during a CBS News interview scheduled to air Sunday morning.

Currently, Anthropic is working to reassure businesses and government agencies that the Trump administration’s supply chain risk label only impacts the use of Claude, their AI assistant and coding tool, when defense contractors employ it for Pentagon projects.

“Your use for any other purpose is unaffected,” Anthropic explained in their statement.

This clarification is vital for Anthropic since most of their anticipated $14 billion in annual revenue comes from businesses and government agencies using Claude for programming and other functions. Over 500 clients pay Anthropic at least $1 million yearly for Claude access, according to recent investment announcements valuing the company at $380 billion.

Anthropic’s Claude technology has gained significant momentum, positioning itself as a potential substitute for various business software tools currently offered by major tech corporations like Salesforce and Workday. This potential has caused stocks of business software companies to decline this year.

However, with Anthropic now labeled as a supply chain risk, questions arise about whether customers will remain comfortable using Claude for non-military purposes, potentially risking Trump’s displeasure. Any widespread hesitation to use Claude could slow AI advancement in the United States during a critical period of technological competition with China.

Simultaneously, Anthropic and Amodei may now have a prominent platform to advocate for stronger AI safety measures.

“No amount of intimidation or punishment from the Department of War will change our position on mass domestic surveillance or fully autonomous weapons,” the company declared. “We will challenge any supply chain risk designation in court.”

In his CBS interview, Amodei framed Anthropic’s disagreement with the Trump administration as defending democratic principles.

“Disagreeing with the government is the most American thing in the world,” Amodei stated. “And we are patriots. In everything we have done here, we have stood up for the values of this country.”

Shortly after Anthropic faced punishment, OpenAI’s Altman announced Friday evening that his company reached an agreement to provide AI services to classified military networks. However, Altman revealed that the same AI limitations that caused Anthropic’s dispute with the Pentagon are now included in OpenAI’s new partnership.

In an internal memo obtained by The Associated Press, Altman informed OpenAI staff: “We have long believed that AI should not be used for mass surveillance or autonomous lethal weapons, and that humans should remain in the loop for high-stakes automated decisions. These are our main red lines.”

The reason why the Pentagon accepted OpenAI’s restrictions but rejected Anthropic’s remains unclear. In his memo, Altman suggested the company believes it can “de-escalate things” by collaborating with the Pentagon while maintaining proper safety protections.

OpenAI’s agreement with the Trump administration coincided with announcing another $110 billion funding round, bringing the San Francisco company’s valuation to $730 billion.

OpenAI may also face potential criticism if consumers who use ChatGPT view its Pentagon work as prioritizing profits over AI safety.

The Anthropic conflict could create new opportunities for Elon Musk, who co-founded OpenAI with Altman in 2015 before their bitter separation over safety and financial disagreements. Musk has accused Altman of fraudulent behavior in a case set for trial in late April.

Musk now operates the AI chatbot Grok, which the Pentagon also plans to connect to classified military networks despite safety and reliability concerns, plus government investigations into its creation of inappropriate deepfake images. Musk has already supported the Trump administration’s conflict with Amodei, posting on his X platform that “Anthropic hates Western Civilization.”

Google, which has created widely-used AI tools through its Gemini technology, could also compete for additional military contracts, though some employees have urged executives to avoid deals that would contradict the company’s former “Don’t be evil” principle. Google leadership has not yet publicly addressed Anthropic’s dispute with the Trump administration.