Anthropic Pulls Top AI Models After U.S. Restricts Foreign Access
Anthropic has abruptly suspended access to two of its most powerful artificial intelligence models after the Trump administration moved to block their use by foreign governments, companies and individuals.
The company said it had taken Fable 5 and Mythos 5 offline for all users after receiving an export control directive from the U.S. government. The order prohibits access by foreign nationals, including people outside the United States and foreign nationals living or working inside the country.
While the restriction was aimed at foreign access, Anthropic said the practical effect was much broader. Because the company could not immediately guarantee that only eligible U.S. users would retain access, it disabled both models entirely while it works through the compliance implications.
The decision marks one of the most significant interventions yet by Washington in the release and use of frontier AI systems. Until now, much of the U.S. government’s technology control strategy has focused on advanced chips, cloud infrastructure and exports to rival states. The Anthropic order suggests that the models themselves are now being treated as sensitive national security assets.
Fable 5 had been positioned as Anthropic’s most capable widely available model, designed for demanding reasoning, agentic work and complex software tasks. Mythos 5 was understood to be even more tightly controlled, with access limited because of its advanced capabilities, particularly in areas such as cybersecurity research and vulnerability discovery.
The Commerce Department has not publicly detailed the specific national security concerns behind the order. Reports indicate that officials were concerned about the potential for the models to be misused or bypassed in ways that could accelerate cyber operations. Anthropic has said it disagrees with the breadth of the intervention and is seeking to restore access as soon as possible.
The shutdown affects customers globally, including businesses that had begun integrating the models into products, workflows and security tools. Developers relying on Fable 5 or Mythos 5 are now being forced to fall back to other Anthropic models or competing AI systems.
Anthropic has stressed that its other tools remain available. However, the sudden withdrawal of its most advanced models raises broader questions for enterprise users about reliability, vendor dependence and geopolitical risk. For companies building products around frontier AI, the incident is a reminder that access can be shaped not only by pricing and technical performance, but also by national security policy.
The move is also likely to intensify debate over how governments should regulate highly capable AI. Supporters of the restriction are expected to argue that powerful models with dual-use potential should be controlled before they can be exploited by hostile actors. Critics will see the order as a blunt intervention that could disrupt innovation, damage customer confidence and push users towards less transparent alternatives.
The episode places Anthropic in a difficult position. The company has often supported stronger AI safety oversight, but it has also argued that regulation should be transparent, technically grounded and applied consistently across the industry. A sudden export restriction on only two of its models could fuel claims that AI governance is becoming reactive and politically driven.
For the wider AI sector, the message is clear. The race to build more powerful models is no longer just a competition between technology companies. It is becoming a matter of trade policy, cybersecurity strategy and international power.
If access to frontier models can be switched off almost overnight, businesses and governments outside the United States may now accelerate efforts to diversify their AI suppliers, develop domestic systems or rely more heavily on open-source alternatives.
Anthropic says it is working to resolve the issue and restore access where possible. Until then, Fable 5 and Mythos 5 remain unavailable, and the AI industry has been given a stark preview of how quickly the politics of artificial intelligence can reshape the technology landscape.
Photo Credit: DepositPhotos.com