Anthropic to disable its most advanced AI models after US order limiting foreign access | AI (artificial intelligence)

Anthropic is disabling its Fable 5 and Mythos 5 AI models for all users following a US government export control directive. The Commerce Department ordered the suspension of access for foreign nationals due to national security concerns regarding potential “jailbreaks” that could identify software vulnerabilities, according to company statements and government officials.

Why is the US government blocking Anthropic’s Fable 5 and Mythos 5?

The US government issued an export control directive to suspend access to these specific models for all foreign nationals. According to a statement from Anthropic, the government believes a method exists to “jailbreak” safeguards that prevent Fable 5 from being used to identify software vulnerabilities.

Anthropic says it received only “verbal evidence of a potential narrow, non-universal jailbreak” and disagrees that this should lead to recalling a commercial model used by hundreds of millions of people. To ensure full compliance with the directive, the company is abruptly disabling both Fable 5 and Mythos 5 for its entire user base.

How does this clash reflect tensions between AI labs and regulators?

The shutdown highlights a growing rift between AI developers and the US government over risk assessment. While Anthropic recently called for greater oversight and the ability to block high-risk models, the company claims this specific action doesn’t follow “fair and fact-based regulation.”

This isn’t the first friction point. Anthropic’s relationship with the government soured earlier this year after the company refused to let the US military use its models for autonomous weapons and domestic surveillance. In response, the government placed Anthropic on a supply chain blacklist, which takes effect later this year.

Kirsten Davies, the Pentagon’s chief information officer, defended the priority of security over profit in a post on X. Davies stated that “some things are simply more important than revenue cycles, clickbait and pre-IPO valuation,” adding, “America First. Always.”

What are the risks associated with “Mythos-class” AI models?

Anthropic’s “Mythos-class” tier, which includes Fable 5, represents a significant jump in capability. While the company built in guardrails to block cybersecurity risks, some users have called these restrictions “overly broad.”

Experts warned that these models could accelerate sophisticated cyber-attacks if they fall into the wrong hands. They specifically cited the banking sector, which often relies on complex, interconnected legacy technology systems that are vulnerable to the kind of bug-hunting capabilities these models possess.

Will this impact Anthropic’s IPO and market position?

The timing is precarious for Anthropic, which confidentially filed for a US IPO last month. The company is currently racing rival OpenAI to reach public markets.

The government’s directive and the existing supply chain blacklist create a volatile environment for potential investors. While Anthropic claims it is working to restore access and believes there is a “misunderstanding,” the abrupt loss of its most advanced models could affect its valuation and user growth metrics.

Frequently Asked Questions

Are all Anthropic models disabled?
No. Anthropic stated that access to all other models remains unaffected; only Fable 5 and Mythos 5 are disabled.

What is an AI “jailbreak”?
A jailbreak is a method used to bypass a model’s built-in safety guardrails, allowing it to generate content or perform actions it was designed to block, such as identifying software vulnerabilities.

Who issued the order to suspend the models?
A US official confirmed the directive was issued by the Department of Commerce.