Trump Administration Moves to Ban Anthropic AI from Federal Government

Presidential Order Shakes Up Defense Tech Landscape

In a dramatic escalation of tensions between Silicon Valley and the Pentagon, President Trump issued a sweeping directive on Friday ordering all federal agencies to immediately stop using technology from Anthropic, a leading artificial intelligence company. Taking to Truth Social, the president didn’t mince words, declaring “We don’t need it, we don’t want it, and will not do business with them again!” The announcement represents one of the most significant confrontations yet between the federal government and an AI company, raising fundamental questions about who controls how powerful AI systems are used by the military. While most agencies face an immediate ban, the Department of Defense has been given a six-month window to transition away from Anthropic’s Claude AI model. The president’s post carried an unmistakable warning: the company had better cooperate during this phase-out period, or face “the Full Power of the Presidency” with potential “major civil and criminal consequences to follow.” Trump characterized Anthropic as a “Radical Left AI company run by people who have no idea what the real World is all about,” setting a combative tone that signals a broader ideological battle over AI governance.

Defense Secretary Declares AI Company a Security Risk

Approximately ninety minutes after President Trump’s explosive announcement, Defense Secretary Pete Hegseth made good on previous threats by officially designating Anthropic as a supply chain risk to national security. In his own strongly-worded statement, Hegseth announced that effective immediately, no contractor, supplier, or partner doing business with the U.S. military would be permitted to conduct any commercial activity with Anthropic. The Defense Secretary, who has rebranded his department as the “Department of War,” framed the decision in stark terms: “America’s warfighters will never be held hostage by the ideological whims of Big Tech. This decision is final.” Despite the harsh rhetoric and the supply chain designation, Hegseth’s directive does allow Anthropic to continue providing services to the military for up to six months to ensure what he called “a seamless transition to a better and more patriotic service.” The language used by both Trump and Hegseth suggests this conflict extends beyond mere contractual disagreements into deeper questions about corporate responsibility, national security, and the appropriate limits on military technology.

The Heart of the Dispute: AI Guardrails and Military Use

At the center of this high-stakes standoff lies a fundamental disagreement about safeguards on artificial intelligence used for military purposes. Anthropic, led by CEO Dario Amodei, has been pushing for specific restrictions on how the Pentagon uses its Claude AI model, while military officials have insisted on unrestricted access for “all lawful purposes.” According to sources familiar with the negotiations, Anthropic requested two primary safeguards: first, that Claude not be used to conduct mass surveillance of American citizens, and second, that the AI system not make final targeting decisions in military operations without human oversight. The company’s concerns are rooted in technical reality—like all current AI systems, Claude is susceptible to what experts call “hallucinations,” where the AI generates incorrect or nonsensical outputs. In a military context, such errors could potentially lead to catastrophic consequences, including unintended escalation of conflicts or mission failures that cost lives. The company argues that human judgment must remain in the loop for critical decisions. The Pentagon had set a hard deadline of 5 p.m. Friday for Anthropic to drop these guardrails and had even threatened to invoke the Defense Production Act—a Cold War-era law that gives the government power to compel private companies to prioritize national defense needs—to force compliance.

The Business Relationship and Its Implications

The dispute carries significant stakes for both parties. Last July, Anthropic was awarded a $200 million contract from the Pentagon to develop AI capabilities advancing national security objectives. More importantly, Anthropic currently holds a unique position as the only AI company with its model deployed on the Pentagon’s classified networks, achieved through a partnership with Palantir, the data analytics company known for its deep government relationships. This exclusive access makes the breakdown in relations particularly consequential. Pentagon officials have suggested alternatives exist, noting that Grok—the AI model owned by Elon Musk’s xAI company—could potentially be deployed in classified settings. This mention is particularly notable given Musk’s close relationship with the Trump administration and his increasingly prominent role in government technology discussions. Pentagon sources told CBS News that their concern extends beyond this specific contract dispute: they argue that a contractor who believes it has veto power over government policy decisions fundamentally cannot be relied upon to work cooperatively with other U.S. partners and contractors. From the military’s perspective, this isn’t just about one AI company’s preferences—it’s about establishing clear lines of authority over defense technology.

Competing Claims About Compromise Efforts

Both sides have offered sharply different accounts of negotiations leading up to the final breakdown. Emil Michael, the Pentagon’s chief technology officer, told CBS News on Thursday that the military had “made some very good concessions” to reach a deal with Anthropic. According to Michael, the Defense Department offered to put in writing specific acknowledgments of federal laws restricting military surveillance of Americans and included language recognizing existing Pentagon policies on autonomous weapons. “At some level, you have to trust your military to do the right thing,” Michael argued, suggesting that Anthropic’s insistence on contractual safeguards implied a lack of faith in military judgment and existing legal frameworks. However, an Anthropic spokesperson painted a very different picture of these negotiations. The company stated that the new contract language received from the Pentagon “made virtually no progress on preventing Claude’s use for mass surveillance of Americans or in fully autonomous weapons.” More troublingly from Anthropic’s perspective, they claimed the proposed “compromise” language was deliberately undermined by legal provisions that “would allow those safeguards to be disregarded at will.” In his own statement, CEO Amodei made clear that threats from the Defense Department would not change the company’s position on needed guardrails, though he expressed Anthropic’s strong preference to continue serving military personnel with appropriate safeguards in place.

Political Fallout and Broader Implications

The Trump administration’s aggressive move against Anthropic has already generated significant political controversy. Democratic Senator Mark Warner of Virginia, who serves as vice chair of the Senate Intelligence Committee, issued a sharp rebuke, accusing the president and defense secretary of “bullying” the company to deploy “AI-driven weapons without safeguards”—something he said “should scare the hell out of all of us.” Warner raised concerns that extend beyond this single dispute, questioning whether national security decisions are being “driven by careful analysis or political considerations.” His statement reflects growing unease among lawmakers about how emerging AI technologies are being integrated into military operations and who gets to set the rules governing their use. This confrontation highlights fundamental tensions that will only intensify as artificial intelligence becomes more capable and more deeply embedded in military systems. On one side stands the argument for democratic accountability and civilian control—the principle that elected officials and military leaders, not private companies, should determine how defense capabilities are employed. On the other side is the concern that AI systems, despite their impressive capabilities, remain imperfect tools that require carefully designed limitations to prevent catastrophic errors. The resolution of this dispute will likely establish important precedents for how the United States military acquires and deploys AI technology in the years ahead, with implications extending far beyond a single contract with a single company. As AI systems become increasingly central to national security, the question of who controls their deployment and under what constraints will remain one of the most consequential policy debates of our time.