A Democracy Drive Thread
How the administration tried to force an AI company to drop its safety limits — and what it did when the company refused.
February 18, 2026
Defense Secretary Pete Hegseth met Anthropic CEO Dario Amodei and gave the company a Friday deadline to allow the military unrestricted use of its Claude models. Anthropic’s stated red lines were narrow but firm: it would not let its AI be used for autonomous weapons or for mass domestic surveillance of Americans.
February 24, 2026
When Anthropic held its line, Hegseth threatened to either declare it a supply-chain risk or invoke the 1950s Defense Production Act to compel access. He and other officials derided the company’s safety stance as “woke AI,” called Anthropic “sanctimonious,” and accused it of trying to “strong-arm” the U.S. military. Elon Musk posted that “Anthropic hates Western Civilization,” which Hegseth amplified; Trump branded the firm an “out-of-control, Radical Left AI company.”
March 5, 2026
The Pentagon formally designated Anthropic a “supply chain risk” — a label normally reserved for firms tied to foreign adversaries — effective immediately, after the company again refused to remove its guardrails. The designation functions as a government blacklist, barring military contractors from doing business with the company.
March 9, 2026
Anthropic filed two federal lawsuits against the Trump administration, alleging the Pentagon illegally retaliated against it for its AI-safety positions and warning that the designation would harm U.S. competitiveness in AI.
June 12, 2026
The Commerce Department invoked export-control rules to bar Anthropic from making Fable 5 and its underlying Mythos 5 model available to any foreign national — inside or outside the U.S., including Anthropic’s own foreign-national employees. Rather than comply piecemeal, Anthropic pulled both of its most powerful systems from public access. The trigger was a security finding raised on a call involving Amazon CEO Andy Jassy, after Amazon researchers prompted Fable 5 into producing information that could aid cyberattacks. It was the first time the federal government had forced a leading AI company to withdraw its systems from public use.
June 14, 2026
Two days after Washington forced Fable 5 and Mythos 5 offline, Hegseth publicly celebrated the Pentagon’s break with Anthropic, posting on X that the renamed “Department of War” had expelled the company from its building months earlier — and that the passage of time only vindicated the move. The department said it had since shifted much of its AI work off Anthropic’s models.
“Three months ago, @DeptofWar kicked @AnthropicAI out of our building—forever. Every passing day proves why that was the right move.”