Anthropic announced this Friday the suspension of public access to its most advanced artificial intelligence models, Fable 5 and Mythos 5, after receiving an export control order from the Trump Administration.
The directive obliges the company to prevent foreign individuals from accessing these services for reasons of national security, whether they are outside or inside the United States.
Given the impossibility of immediately implementing a selective block, Anthropic opted to cut off access to all its clients, including paying users.
The last-minute order
As the company explained, the US government's communication arrived on Friday afternoon. The company assured that, in order to comply with the last-minute directive, it decided to generally suspend access to Fable 5 and Mythos 5 while it evaluates how to adapt the service to the new requirements.
Anthropic apologized to its clients for the interruption and acknowledged that the measure affects users who had contracted access to its most advanced tools.
Anthropic's criticisms of the procedure
In its statement, Anthropic stated that the US government has not provided it with specific details about the technical reasons why its new models pose a national security problem.
The company argued that the Executive Branch should have the ability to block insecure developments, but insisted that this process should be transparent, fair, clear, and based on technical facts.
Anthropic maintained that the decision received this Friday does not meet these principles, as it was applied without sufficient public explanation and without a clear framework for the company and its clients.
Fable 5, the company's most advanced model
The suspension comes just days after Anthropic made the Fable 5 model available to its Claude subscribers, presented as its most advanced system to date.
The company had highlighted that Fable 5 surpassed its Opus models in various performance and capability tests, positioning it as one of its main bets in the race for next-generation artificial intelligence models.
The connection with Mythos
Fable 5 was based on the Mythos model, a system that had already generated concern within the company itself.
Anthropic warned in April that Mythos presented particularly sensitive capabilities, with potential risks to the security of advanced cybersecurity systems in critical sectors, including the financial sector.
For this reason, the company previously allowed the US Government and several companies to analyze Mythos before its public deployment, with the aim of identifying vulnerabilities and assessing their impact on security.
Safeguards and restrictions
Anthropic maintains that Fable 5 incorporated safeguards and restrictions intended to prevent misuse. The company had designed the model with specific limitations to reduce risks associated with advanced capabilities, especially in sensitive areas such as cybersecurity.
Even so, the US Administration has decided to activate the export control directive, which has led to the immediate suspension of public access.
Project Glasswing
Although Anthropic assured that it would not publish the Mythos model openly, it did make it available to a very small list of companies within the cybersecurity initiative Project Glasswing.
That program sought to allow selected actors to evaluate the model's potential in controlled environments, especially to detect security flaws and strengthen defenses against advanced threats.
The clash between AI and national security
The suspension of Fable 5 and Mythos 5 opens a new episode in the debate over the control of the most advanced artificial intelligence models.
The case reflects the growing tension between technological innovation, commercial interests, national security, and public regulation, especially when models can have sensitive applications in strategic sectors or capabilities that are difficult to control once deployed.