US Government Orders Anthropic to Disable Claude Fable 5 and Mythos 5 Globally

The US government has ordered Anthropic to restrict access to its newest and most powerful AI models, Claude Fable 5 and Claude Mythos 5, resulting in a complete worldwide shutdown of both systems. The directive, issued late Friday, came just days after the public launch of Fable 5 and has sent ripples through the global artificial intelligence community.

Anthropic, the San Francisco headquartered company backed by major investors including Amazon and Google, issued a detailed public statement confirming the suspension. According to the company, the US Commerce Department instructed it to block all access for foreign nationals. This includes individuals located inside the United States as well as Anthropic’s own foreign born employees. Faced with strict “deemed export” compliance requirements, Anthropic concluded that temporarily disabling the models for every user was the only viable option.

The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance,” Anthropic said. The company added that other existing Claude models, such as the Opus 4 series, continue to operate normally. Anthropic described the government action as based on a misunderstanding and stated that it is working closely with officials to resolve the issue and restore access quickly.

Details on the Affected Models

Claude Fable 5 made its debut on June 9 as the first publicly available model in Anthropic’s new Mythos class series. Early feedback from developers highlighted its strong performance in complex coding projects, agentic task handling, extended context reasoning reaching up to one million tokens, and specialized cybersecurity applications. Many users noted its ability to manage large scale software engineering work that would typically require weeks of coordinated team effort.

Fable 5 had also recently demonstrated its effectiveness in security research, with reports indicating that the model successfully identified and helped prevent a potential Ethereum USDT exploit scenario during testing. The achievement strengthened confidence among developers who viewed the model as a practical tool for both software development and cybersecurity applications.

The more advanced Claude Mythos 5 had been available exclusively to a limited group of trusted partners under the Project Glasswing initiative. This version included enhanced capabilities with comparatively lighter restrictions in certain high sensitivity areas. Together, these models were seen as important advancements over previous Claude releases, offering improved balance between raw power and responsible safety measures.

The capabilities of the Mythos series had already attracted significant attention within the AI and cybersecurity communities. Anthropic previously showcased the technology through its Project Glasswing program, where selected partners were given early access to evaluate advanced cybersecurity and reasoning capabilities before wider deployment. This limited preview helped establish Mythos 5 as one of the company’s most anticipated AI releases.

Reasons Behind the Government Action

The Commerce Department directive reached Anthropic around 5:21 PM ET on June 12. It classifies access by non US persons as a controlled export that requires specific licensing. Sources indicate the primary concern involves a reported jailbreak technique that might allow bypassing of safety alignments, especially in areas related to cybersecurity. Anthropic has pushed back on the severity, noting that the issue is comparable to similar vulnerabilities identified in other leading AI systems and that it received only limited information about the exact risks involved.

This development aligns with the current Trump administration’s broader strategy of applying export style controls to advanced AI technologies, treating them with the caution reserved for sensitive military assets. Anthropic has experienced prior points of friction with US defense related entities, which provides additional context for the latest regulatory intervention.

Industry Reactions and Global Consequences

Developers and enterprise customers who had begun incorporating Fable 5 into live workflows expressed immediate disappointment and disruption. Discussions on social media platforms showed a mix of criticism toward perceived regulatory overreach and renewed emphasis on the importance of building independent AI capabilities in regions such as India and Europe. Several technology leaders highlighted the event as further proof of the need for stronger national AI strategies outside heavy US influence.

The debate comes amid growing concerns over technology security and geopolitical competition. Recent actions by US authorities, including the seizure of multiple websites allegedly linked to foreign intelligence recruitment efforts, have reinforced Washington’s focus on protecting sensitive technologies and restricting access to capabilities viewed as strategically important.

The timing presents a notable challenge for Anthropic, which is preparing for a potential high profile initial public offering later this year. Although only the newest models are affected, the incident draws attention to the increasing regulatory uncertainties facing frontier AI companies and the practical difficulties of serving a global customer base under tightening security rules.

Experts view this as part of an ongoing pattern in which governments attempt to limit the spread of powerful AI tools to strategic rivals. At the same time, observers caution that overly broad restrictions risk slowing down innovation and encouraging development in less regulated environments.

Anthropic has recommended that users temporarily rely on other available models such as Claude Opus 4.8 while the situation is addressed. The company expects to release further technical details shortly and remains engaged in discussions with relevant authorities.

This case brings into focus one of the central challenges in the current AI era: finding the right equilibrium between protecting national security interests and fostering the kind of open innovation that drives technological leadership worldwide. The resolution of this matter may influence future approaches to regulating and distributing advanced AI systems.