Trump Administration Imposes Export Ban on Anthropic’s Latest AI Models

On Friday, the Trump administration issued an export-control order that forced Anthropic to shut down its two newest models, Fable 5 and Mythos 5, citing unspecified national-security concerns. Anthropic was given roughly 90 minutes to halt access, affecting customers and employees, many of whom are foreign nationals.

TechCrunch

Anthropic’s friction with Washington began when the Pentagon labeled the company a “supply chain risk” and demanded guard-rail changes for military deployments. Amazon CEO Andy Jassy raised a reported jailbreak—first flagged by Amazon researchers—to the White House, prompting the ban. Legal scholar Jessica Tillipman later criticized the lack of a clear process.

CNN

Fable 5 launched June 9, 2026; the Commerce Department issued the export-control order two days later; Anthropic disabled both models on June 13. President Donald Trump told the G7 summit on June 21 that talks were “going fine.”

CNN

Two models—Fable 5 (public) and Mythos 5 (partner preview)—were banned. Anthropic, valued near $1 trillion, says the jailbreak is minor and also appears in other models.

CNN

The administration called the models a “national security risk” after a reported jailbreak. Anthropic argued the flaw is not unique and does not merit the ban. Trump said talks were proceeding well, then later softened his threat.

CNN

Dozens of cybersecurity experts and AI entrepreneurs signed an open letter warning that pulling tools from defenders “is dangerous.” Alex Stamos said the findings do not justify a reaction; David Sacks, former White House AI adviser, called the jailbreak “serious.” Brad Carson warned the ad-hoc approach could set a regulatory precedent.

CNN

The government gave no specific security rationale; Anthropic says the issue is minor, while some experts deem it serious.

CNN

The case exposes the lack of a transparent AI-risk framework, risking both innovation and security.

CNN

• “Anthropic has not had the best relationship with the Trump administration in a way that stands apart from the other leading AI labs,” — Sean O’Kane, TechCrunch host

TechCrunch

• “The problem is not that the government exercised discretion; national security demands such latitude,” — Jessica Tillipman, George Washington University

CNN

• “Right now, you have an ad hoc, personalized, opaque, possibly lawless approach,” — Brad Carson, Public First

CNN

• “There were some valid findings but no unique capabilities that justify a reaction close to this,” — Alex Stamos, AI-safety researcher

CNN

Anthropic and the administration remain in private talks to lift the export ban, with Trump saying progress is being made at the G7 summit. Observers expect the dispute to accelerate the administration’s broader AI policy, potentially yielding a sector-specific regulatory framework. Stakeholders continue to call for a transparent, scientific process for future AI risk assessments.

CNN