Artificial intelligence company Anthropic has announced it will disable access to its newest and most advanced AI models after receiving a directive from the United States government requiring the suspension of access for foreign nationals.

The order affects Anthropic’s recently launched Fable 5 and Mythos 5 models, which the company describes as representing a new generation of AI capability. According to Anthropic, the government instructed it to halt access to the models for all non-US citizens, citing national security concerns, although the company said it was not provided with detailed evidence supporting the decision.

Anthropic said it believes the action stems from concerns that users may have discovered a way to circumvent safeguards designed to prevent Fable 5 from being used to identify software vulnerabilities. The company characterised the issue as a limited and highly specific security concern rather than a widespread flaw.

The decision marks a significant expansion of US efforts to control access to advanced artificial intelligence technologies. Historically, export controls have focused on restricting access to semiconductors, computing infrastructure and AI development tools. The latest directive instead targets direct access to advanced AI models themselves.

The move comes during a turbulent period in Anthropic’s relationship with the US government. Tensions escalated earlier this year after the company reportedly declined to allow its technology to be used for domestic surveillance activities and fully autonomous weapons systems. In response, government agencies placed Anthropic on a supply-chain blacklist that is expected to take effect later this year.

Anthropic criticised the latest action, arguing that the evidence presented did not justify removing access to models already being used by hundreds of millions of people. The company said regulators had only provided verbal information regarding what it described as a narrow and non-universal method of bypassing model protections.

The dispute highlights a growing divide between AI developers and policymakers over how risks associated with advanced models should be evaluated. While Anthropic has previously supported stronger oversight of artificial intelligence and has publicly advocated for regulatory frameworks capable of blocking unsafe models, it said the latest intervention did not reflect evidence-based regulation.

US Department of Defense Chief Information Officer Kirsten Davies defended the government’s position, stating that national security considerations must take precedence over commercial interests and financial objectives.

Anthropic staff to meet White House officials next week, Axios reports | Reuters

Photo by Reuters

The timing is particularly significant for Anthropic, which confidentially filed paperwork for a US initial public offering last month. The company has been widely viewed as one of the leading contenders in the race to become the next major publicly listed AI company, alongside rival OpenAI.

Anthropic only recently introduced Fable 5 as part of a new category of systems it calls Mythos-class models. The company said these models include safeguards that restrict their use in sensitive areas such as cybersecurity, although some users have criticised those restrictions as being overly broad.

Security experts have warned that increasingly capable AI systems could potentially accelerate sophisticated cyberattacks if misused, particularly against industries such as banking and financial services that rely on complex legacy technology infrastructure.

Anthropic said it worked closely with government agencies and external safety organisations prior to the launch of Fable 5 and noted that competing AI models from other providers demonstrate similar abilities when analysing software code and identifying minor defects.

To comply with the directive, Anthropic said it must immediately disable Fable 5 and Mythos 5 across its platforms. Other AI models offered by the company will remain available.

The company maintains that the situation is based on a misunderstanding and said it is actively engaging with government officials in an effort to restore access. Anthropic also warned that applying the same standard across the wider AI industry could significantly disrupt future model launches and slow innovation among leading AI developers.

Amazon Web Services later confirmed that Anthropic had requested the removal of access to the affected models for users across all regions. A US government official subsequently confirmed that the Commerce Department had issued an export control directive restricting access to Fable 5 and Mythos 5 by foreign nationals.

The order could have far-reaching implications for the global AI industry. Former White House adviser Dean Ball suggested the restrictions may ultimately require users to verify their citizenship before gaining access to certain advanced AI systems.

Questions also remain about how the directive could affect employees within AI companies themselves. Several prominent Anthropic researchers and executives were born outside the United States, although the company has declined to comment on whether the restrictions would apply to staff members based on citizenship status.

As governments around the world grapple with the rapid advancement of artificial intelligence, the Anthropic case may become a landmark example of how national security concerns increasingly intersect with access to cutting-edge AI technologies.

Featured photo by Reuters