Pentagon chief slams Anthropic 'betrayal' in AI tech row
WASHINGTON, United States — US Defense Secretary Pete Hegseth accused AI startup Anthropic of "arrogance and betrayal" after it rejected the Pentagon's demand for unconditional military use of its Claude models, emphasizing that American warfighters won't be held hostage by Big Tech's ideological whims.
The dispute between the Pentagon and AI startup Anthropic centers on military use of Anthropic's Claude AI models. The Pentagon issued a Friday deadline (February 28, 2026) demanding that Anthropic allow "all lawful use cases" of Claude on classified networks, threatening to blacklist the company as a "supply chain risk," terminate contracts worth up to $200 million, and potentially invoke the Defense Production Act if their demands weren't met. Pentagon officials, including CTO Emil Michael, have sharply criticized Anthropic, with Michael accusing the company of "lying" on social media about the Pentagon's intentions.
Anthropic CEO Dario Amodei rejected the Pentagon's demands in a Thursday statement, citing ethical "red lines" against using Claude for mass surveillance of Americans or fully autonomous weapons without human intervention. Amodei argued that current AI technology cannot safely handle such scenarios and stated the company "cannot in good conscience accede" to the Pentagon's demands. The company emphasized its commitment to AI safety and ethical guidelines that restrict certain military applications.
Pentagon officials countered that they seek flexibility for legitimate defense applications like countering drone swarms without requiring per-use approvals. They denied interest in illegal mass surveillance of Americans or developing fully autonomous weapons without human oversight. The military argued that Anthropic's restrictions were unnecessarily limiting and could hinder national security operations. Replacing Anthropic's tools, which were cleared for classified systems ahead of competitors, could take three months or longer, potentially disrupting Pentagon operations.
The dispute has drawn attention from other AI industry leaders and lawmakers. OpenAI CEO Sam Altman publicly supported Anthropic's limits, noting shared concerns about military AI use and revealing that his company is engaged in similar negotiations with the Pentagon. Bipartisan Senate leaders, including Armed Services Chair Roger Wicker (R-MS) and Ranking Member Jack Reed (D-RI), sent a private letter urging both sides to resolve the issue. Senator Mark Warner (D-VA) criticized the Pentagon for ignoring AI governance frameworks endorsed by White House offices. As of the deadline passing late Friday, sources indicated no confirmed resolution, with discussions potentially involving a six-month phase-out period for existing systems.
Join the discussion
What do you think? Drop your thoughts below.