U.S. President Donald Trump has instructed every federal department to discontinue reliance on Claude and additional offerings from Anthropic, intensifying a tense dispute between the Defense Department and the firm concerning protections for artificial intelligence. In a Friday afternoon message on Truth Social, Trump announced a six-month transition timeline for agencies, encompassing the Defense Department, to shift away from Anthropic's technologies.
Trump accused Anthropic's 'left-wing extremists' of committing a severe error by attempting to dictate terms to the military branch, prioritizing their policies over constitutional principles. He warned the company to cooperate during the transition or face the complete authority of the executive branch, potentially leading to significant legal and penal repercussions.
Prior to this announcement, Defense Secretary Pete Hegseth had warned of classifying Anthropic as a potential threat in the supply chain unless it removed restrictions preventing Claude's application in widespread monitoring of U.S. citizens or in weapons operating without human oversight. Following Trump's remarks, Hegseth posted on X, stating he was ordering the military to mark Anthropic as a national security supply-chain hazard, barring any contractors, vendors, or affiliates working with the U.S. armed forces from engaging in transactions with the company starting right away.
Anthropic offered no immediate reply to inquiries from Engadget. Earlier that day, a company representative noted that the agreement proposed after CEO Dario Amodei detailed Anthropic's stance showed almost no advancement in curbing the specified improper applications.
"The proposed adjustments, presented as concessions, came with fine print permitting those protections to be ignored whenever desired. Contrary to the military's latest declarations, these limited measures have formed the core of our discussions for several months," the representative explained. "We are prepared to resume discussions and dedicated to ensuring seamless operations for the military and U.S. troops."
Organizations advocating for democratic principles, such as the Center for Democracy and Technology (CDT), swiftly opposed the president's warnings. "Such measures establish a risky benchmark. They deter corporations from openly debating suitable applications of their innovations with authorities, a vital aspect in security contexts with limited transparency," stated CDT leader Alexandra Givens in a comment provided to Engadget. "These warnings erode trust in the creative environment, skew economic drivers, and promote an overly broad interpretation of presidential authority that concerns citizens regardless of ideology."
Currently, support within the artificial intelligence sector appears aligned with Anthropic. On Friday, numerous staff members from Google and OpenAI endorsed a public declaration calling on their employers to show unity with the organization. An internal document reviewed by Axios revealed that OpenAI's chief executive, Sam Altman, affirmed the company would uphold identical boundaries as Anthropic.
In a late Friday blog entry, Anthropic pledged to legally contest any supply-chain risk classification and informed clients that solely activities tied to the Defense Department would face disruption. The complete declaration can be found on the company's site; a portion follows:
"Labeling Anthropic as a supply-chain concern represents an extraordinary step—typically used against foreign opponents, not openly against a domestic enterprise. We are profoundly disappointed by these events. As the initial advanced AI provider to integrate systems into secure U.S. government environments, Anthropic has aided American forces since June 2024 and plans to persist in that role."
"We view this classification as both invalid under law and harmful to future precedents for U.S. firms dealing with officials."
"No level of pressure or penalties from the military will alter our stance against large-scale internal monitoring or independent lethal systems. We will pursue judicial review of any such classification."