Anthropic has suspended access to its newest AI systems after receiving a directive from the United States government, marking one of the most unusual interventions yet in the fast-growing artificial intelligence sector. The company, known for its Claude chatbot, confirmed that it had to disable its latest models, Fable 5 and Mythos 5, following national security instructions issued late Friday.
According to Anthropic, the U.S. government ordered the company to prevent any foreign nationals from accessing its most advanced AI models. The restriction applied not only to users outside the United States but also to foreign employees within the company. As a result, Anthropic said it was forced to fully suspend access to both models for all customers to ensure compliance.
The company described the directive as an export control measure linked to national security concerns. It said the instruction came from the Commerce Department and was supported by officials from the Bureau of Industry and Security. The letter was reportedly sent by Commerce Secretary Howard Lutnick, according to a U.S. administration official.
Anthropic said the decision to shut down access was difficult and described it as the first known case where a major AI company has taken publicly released models offline due to direct government intervention. On its platform, users saw messages indicating that Fable 5 was temporarily unavailable.
The company explained that the government raised concerns about a specific technical weakness that could allow users to bypass safety protections, a method often referred to as a jailbreak. Officials reportedly argued that this issue could be exploited by foreign nationals, raising national security risks.
However, Anthropic said it was only provided with verbal evidence of the issue and argued that the concern did not justify removing widely deployed models. The company also said the same vulnerability could exist in other AI systems across the industry.
Anthropic strongly disagreed with the decision, stating that a limited safety flaw should not lead to the removal of commercial models used by millions of people. It also suggested that the directive may have been based on a misunderstanding of the technical risk involved.
The company had only recently launched Fable 5 and Mythos 5. The models were presented as its most powerful AI systems to date, with improved reasoning and advanced capabilities. Anthropic also warned at the time of release that such systems carry risks if misused, particularly in sensitive areas like cybersecurity.
Fable 5 was made available to the general public but with strict safety restrictions, especially around topics such as cyber operations and biological information. Mythos 5, by contrast, was released to a smaller group of trusted partners, including cybersecurity and infrastructure organisations, and included fewer restrictions.
The sudden suspension highlights growing tension between AI companies and governments over how powerful models should be controlled. It also raises questions about how national security rules should apply to rapidly advancing AI systems that are widely accessible across borders.
The situation is also part of a broader and sometimes strained relationship between Anthropic and the U.S. government. Earlier this year, the company faced restrictions from federal agencies after disagreements over AI safety policies for defence use. That dispute later moved into the courts and remains ongoing.
Despite previous tensions, there had recently been signs of cooperation between the two sides. Reports suggested that U.S. security agencies had begun using Anthropic’s AI tools for cyber defence and offensive operations, reflecting the growing importance of AI in national security strategies.
In June, the U.S. administration also issued an executive order focused on artificial intelligence. It directed federal agencies to strengthen cyber defence systems and explore ways to gain early access to advanced AI models developed by leading companies.
Anthropic said it supports transparent and structured government oversight but argued that the latest directive did not meet those standards. The company expressed hope that access to Fable 5 and Mythos 5 could be restored once the issue is reviewed more clearly.
For now, the suspension places some of the most advanced AI tools in the industry out of public reach, underscoring the rising regulatory pressure surrounding frontier AI development and its potential security risks.
