Federal Retaliation Against Anthropic Over AI-Use Limits (2026)
In early 2026 the Trump administration moved to punish the artificial-intelligence company Anthropic after it refused, in contract negotiations, to give the Pentagon unrestricted access to its Claude models. Anthropic's acceptable-use policy barred its technology from powering fully autonomous weapons and the mass domestic surveillance of Americans; the Department of Defense demanded access "for all lawful purposes," and when the company declined, the administration retaliated rather than negotiating the terms. On February 27 President Trump ordered federal agencies and military contractors to stop doing business with Anthropic and publicly branded the company's stance a "DISASTROUS MISTAKE," and on March 5 Defense Secretary Pete Hegseth formally designated Anthropic a "supply chain risk" — a label normally reserved for firms tied to foreign adversaries and reportedly the first applied to an American company — even as the military kept using Claude for intelligence and targeting work. Anthropic called the designation unprecedented and legally unsound and sued the Defense Department on March 9. This episode collects The Standing's entries on the administration's campaign to penalize a private company for the limits it placed on how its technology may be used.
Documented in this episode (3)
Trump orders federal agencies and contractors to cut off Anthropic over its AI-use limits
On February 27, 2026, President Trump ordered federal agencies and military contractors to halt business with the AI company Anthropic, giving the government roughly six months to phase out its products. The order followed the breakdown of Pentagon contract talks: the department sought access to Anthropic's Claude models "for all lawful purposes," while the company's acceptable-use policy barred their use for fully autonomous weapons and the mass domestic surveillance of Americans. Announcing the move on Truth Social, Trump called the company's refusal a "disastrous mistake," opening a retaliation campaign that culminated in the March 5 "supply chain risk" designation.
Pentagon brands Anthropic a 'supply chain risk' in retaliation for refusing unrestricted military use of its AI models
On March 5, 2026, the U.S. Department of Defense formally notified Anthropic that the company and its products were designated a "supply chain risk," effective immediately — a label normally reserved for firms tied to foreign adversaries and reportedly the first publicly applied to an American company. The designation, which bars defense contractors from using Anthropic's technology, followed the breakdown of contract talks: the Pentagon demanded access to Anthropic's Claude models "for all lawful purposes," while the company's acceptable-use policy barred their use for fully autonomous weapons and mass domestic surveillance of Americans. After Anthropic refused to drop those limits, President Trump on February 27 ordered agencies and contractors to halt business with the company and called its stance a "disastrous mistake," and Defense Secretary Pete Hegseth tied the designation directly to the firm's refusal to comply — even as the military continued using Claude to support intelligence and targeting work.
Commerce Department forces Anthropic to disable two AI models for all foreign nationals in a first-of-its-kind export-control order
On June 12, 2026, the U.S. Department of Commerce — in a letter from Secretary Howard Lutnick to Anthropic CEO Dario Amodei — issued an export-control directive citing national-security authorities that suspended all access to Anthropic's Fable 5 and Mythos 5 models by any foreign national, inside or outside the United States, including the company's own foreign-national employees. To comply, Anthropic abruptly disabled both models for all customers worldwide; its other models were unaffected. The government's stated basis was a belief that a method of "jailbreaking" Fable 5 existed, though the letter gave no details; Anthropic said the cited technique surfaced only minor, previously known vulnerabilities also found in other public models and disputed that it justified recalling a model used by hundreds of millions. Contemporaneous reporting described the order as the first time the U.S. government has forced a leading American AI company to take a publicly deployed model offline.