Pentagon Threatens to Cut Off Anthropic Over AI Use Limitations for Military

The Pentagon is pushing major AI companies—Anthropic, OpenAI, Google, and xAI—to make their advanced models fully available for military missions, explicitly including weapons development, intelligence gathering, and battlefield support ^[1][7]. Unlike the other vendors, Anthropic has held firm, negotiating for months to maintain limits on how its Claude model can be used by government clients. According to multiple reports, Anthropic's opposition has focused on two major red lines: use in autonomous weapons and in large-scale domestic surveillance ^[1][3][4][6]. These positions have deeply frustrated defense officials, who are now threatening to terminate Anthropic's relationship if it does not comply ^[3][4]. The standoff is heightened by reports that Claude was already deployed—via Palantir—in the U.S. mission to capture Venezuela’s ex-president Nicolas Maduro, underscoring the operational stakes as both sides clash over AI governance ^[1][4][6][7].