US Government Orders Anthropic to Take Down Top AI Models

In an unprecedented regulatory action, the US government has forced Anthropic to globally disable its two most advanced artificial intelligence models — Claude Fable 5 and Claude Mythos 5 — just three days after their launch, citing national security export control authorities. Anthropic has publicly stated that the decision is based on “serious misunderstandings” and warned that applying the same standard across the industry would “essentially halt all new model deployments for all frontier model providers.”

The Government Directive

On the evening of June 12, 2026, Anthropic received an export control directive from the US government ordering the immediate suspension of access to both models for “any foreign national, whether inside or outside the United States, including foreign national Anthropic employees,” according to Anthropic’s official statement. Because cloud-based AI systems cannot easily segregate access by nationality, the company was forced to implement a global shutdown, affecting all users worldwide — including paying enterprise customers.

Anthropic said the government letter “did not provide specific details of its national security concern.” The company’s understanding is that the government believes it has identified a method of bypassing, or “jailbreaking,” Fable 5’s safety systems.

Anthropic’s Defense

Anthropic disputes the government’s basis for the directive, arguing that the alleged jailbreak is narrow and non-universal. The company stated that the government has only provided “verbal evidence of a potential narrow, non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws.” Anthropic told CNBC that the capabilities demonstrated in the alleged jailbreak are “widely available from other models,” including OpenAI’s GPT-5.5.

The company had conducted extensive pre-launch red-teaming — thousands of hours with the US government, the UK AISI, and multiple private third-party organizations — and maintains that no tester has yet found a “universal jailbreak” capable of broadly bypassing the model’s safeguards. Anthropic adopted a “defense in depth” strategy with Fable 5, combining narrow jailbreak resistance with thorough monitoring to detect and shut down successful attacks.

The Jailbreak Controversy

The government action follows a viral claim by prolific AI jailbreaker “Pliny the Liberator,” who on June 10 published what he described as a successful jailbreak of Fable 5 on X, demonstrating multi-agent bypass techniques. However, as SecurityWeek reported, Anthropic disputes that the demonstrated technique constitutes a true jailbreak of its core safeguards, noting that some outputs were not produced by Fable 5 at all, while those that were contained only publicly available information.

A Pattern of Tensions

This incident marks the second major confrontation between Anthropic and the US government in 2026. In February, the Pentagon — under Secretary of Defense Pete Hegseth — designated Anthropic a “supply chain risk to national security” after the company refused to allow its Claude models to be used for domestic surveillance and fully autonomous weapons systems. The label, historically reserved for foreign adversaries, prohibited defense contractors from using Claude models in military work. Anthropic sued the administration and litigation is ongoing.

Implications for AI Regulation

The incident highlights the absence of a clear statutory framework for AI deployment regulation in the United States. Anthropic CEO Dario Amodei has previously called for FAA-style regulation with transparent, fair, and technically grounded processes. The current action — an opaque export control directive with minimal explanation — represents the opposite approach.

As VentureBeat’s Carl Franzen noted, “the sudden regulatory intervention serves as a stark warning to the enterprise sector: centralized, cloud-based frontier models exist at the absolute mercy of government oversight and vendor compliance.” The incident has triggered widespread discussion about the risks of single-vendor dependency in enterprise AI, with industry analysts calling for supplier diversification and active fallback architectures.

What’s Next

Anthropic says it is actively working with US government officials to clarify misunderstandings and restore access to both models, but no timeline has been provided. The outcome of this standoff could set a critical precedent for how the US government handles future AI safety concerns and whether similar standards are applied to other frontier AI providers such as OpenAI and Google. The industry is watching closely — and so are Anthropic’s customers, who now face the reality that even the most advanced AI systems can be switched off overnight.