Anthropic PBC has suspended all access to its most advanced artificial intelligence models, including Mythos and Fable 5, following a direct order from the Trump administration. The U.S. Commerce Department mandated that these frontier systems be restricted from all foreign nationals, regardless of their location, citing national security concerns regarding potential cybersecurity vulnerabilities. The company complied by shutting off access to all customers globally to ensure adherence to the directive.
Why did the U.S. government restrict AI model access?
The federal government issued the order after identifying that the Fable 5 model could be “jailbroken” to bypass safety guardrails, according to an official statement from Anthropic. The company noted that the administration specifically raised concerns about the model’s ability to conduct cybersecurity tasks. Sources familiar with the matter, speaking on condition of anonymity, confirmed that Amazon Chief Executive Andy Jassy communicated with senior U.S. officials regarding these vulnerabilities before the government imposed the controls.
The U.S. government previously declared Anthropic a “supply-chain risk” earlier this year following disagreements with the Pentagon over the potential use of its technology for military surveillance.
How does this order affect the AI industry?
This directive marks the first time the U.S. government has enforced such sweeping restrictions on the distribution of frontier AI software. While previous administrations limited the export of physical hardware like semiconductors and supercomputers, this move targets the underlying code and model weights. Aidan Gomez, co-founder of the AI startup Cohere, described the move as a “massive wake-up call” for the industry. The policy contrasts with the administration’s own recent executive order, which explicitly stated that it would not pursue a mandatory licensing regime for AI models.
What is the conflict between Anthropic and the administration?
Anthropic has publicly challenged the government’s approach. In a post on its website, the company argued that if a “narrow potential jailbreak” is considered sufficient grounds for recalling a commercial model, it would effectively halt all new deployments across the entire sector. David Sacks, a former Trump AI czar and current co-chair of the President’s Council of Advisers on Science and Technology, countered this on X, stating that the administration remains “bewildered” that Anthropic has not yet remediated the safety issues to restore access.
Comparison: Government Policy vs. Industry Practice
| Policy Area | Government Stance | Anthropic Stance |
|---|---|---|
| Licensing | Mandatory controls on foreign access | Advocates for voluntary pauses |
| Safety | Recall required for jailbreaks | Recalls are disproportionate |
Frequently Asked Questions
Are all Anthropic models currently offline?
Anthropic has suspended access to its most advanced systems, specifically Mythos and Fable 5, to comply with the government order.
Who is affected by the new restrictions?
The order applies to all foreign nationals, whether they are located inside or outside the United States.
Will these models be available again?
According to David Sacks, the administration’s stated goal is for Anthropic to remediate the safety issues so that the export controls can be lifted and the models returned to general release.
The landscape of AI regulation is shifting rapidly. Subscribe to our newsletter for the latest updates on how federal policy impacts the next generation of technology.