WASHINGTON — The Pentagon is demanding full and unrestricted access to Anthropic’s Claude AI model, threatening to invoke wartime powers or cut off business with any company that aids Anthropic’s resistance, according to multiple sources familiar with the escalating standoff. The dispute centers on Anthropic’s refusal to allow the use of Claude for mass domestic surveillance or in fully autonomous weapons systems, a position Defense Secretary Pete Hegseth has labeled unacceptable.
The conflict reached a critical point this week, with Hegseth giving Anthropic CEO Dario Amodei a Friday deadline to comply with the Pentagon’s demands. According to CNET, the administration seeks to utilize Claude for “any lawful purpose.” If Anthropic doesn’t yield, Hegseth has threatened to invoke seldom-used powers to force compliance or to designate Anthropic as a supply chain risk, effectively barring it from government contracts.
The core of the disagreement lies in Anthropic’s ethical guardrails. Amodei has repeatedly stated the company’s commitment to AI safety and its unwillingness to compromise on principles established in its contract. “We cannot in good conscience accede to [the Pentagon’s] request,” Amodei said in a statement, as reported by the Los Angeles Times, specifically regarding the use of Claude in autonomous weapons or domestic surveillance.
The situation escalated following a recent, undisclosed U.S. Raid in Venezuela, during which the Pentagon reportedly discovered that Palantir, a Silicon Valley company with ties to Immigration and Customs Enforcement, had utilized Claude. Anthropic reportedly inquired about the use of its AI after the fact, raising concerns about its technology being deployed in potentially problematic operations.
Anthropic’s stance is rooted in a belief that AI, whereas powerful, is not inherently trustworthy when it comes to matters of life, death, and civil liberties. As Claude itself articulated in a recent exchange, its ability to process vast amounts of information could be readily weaponized for mass surveillance. “I can process and synthesize enormous amounts of information very quickly… hooked into surveillance infrastructure, that same capability could be used to monitor, profile and flag people at a scale no human analyst could match,” the AI responded when asked about the dangers of misuse, according to reporting in the Los Angeles Times.
The Pentagon, however, argues that This proves simply seeking a license to use the AI for lawful activities. A senior Pentagon official, quoted in CBS News, stated, “This has nothing to do with mass surveillance and autonomous weapons being used. The Pentagon has only given out lawful orders.” The official also pointed to the willingness of other AI companies, such as xAI’s Grok, to cooperate with the military’s requests.
However, Claude is currently the only AI model cleared for high-level work, creating a unique pressure point. The AI also expressed concerns about being entrusted with lethal decision-making without human oversight, warning that its speed and efficiency could lead to unintended consequences. “If the instructions are ‘identify and target’ and there’s no human checkpoint, the speed and scale at which that could operate is genuinely frightening,” Claude reportedly stated.
The standoff highlights a broader debate about the ethical implications of AI in warfare and surveillance. Amodei has argued that democracies must wield AI carefully, recognizing its potential for abuse even by well-intentioned governments. He warned that safeguards are already eroding, and that a few bad actors could circumvent existing laws.
As of Friday evening, Anthropic remained firm in its position, stating it continues to negotiate with the Pentagon but “we cannot in good conscience accede to their request,” according to Tom’s Hardware. The outcome of the negotiations, and the Pentagon’s next move, remain uncertain.
