- Anthropic launched Claude Fable 5, its most powerful model yet, on June 9.
- On June 12, US Commerce Secretary Howard Lutnick ordered Anthropic to suspend access to its models for foreign nationals, including employees.
- Due to enforcement issues, Anthropic decided to take the model offline completely for all users.
- At the center of this storm is Mythos, a large language model developed by Anthropic with applications in cybersecurity.
- Mythos was originally developed for a consortium of companies to create critical software for the AI era under Project Glasswing.
- Concerns arose when it was discovered that there was a reliable “jailbreak” to bypass safeguards, prompting the government directive.
- In November 2025, Anthropic detected and disrupted the first largely autonomous, AI-orchestrated cyber espionage campaign attributed to a Chinese state-sponsored group.
- LLMs like Mythos can design new malware that exploits previously unknown vulnerabilities, automating large portions of cyberattacks.
- Since 2022, the US Bureau of Industry and Security has progressively imposed controls on advanced AI chips, citing military application concerns.
- The move to restrict Claude Fable's access is part of a broader trend in government responses to capable AI models.
- The Fable episode may indicate how governments will respond to increasingly capable AI models.
Anthropic's recent launch of Claude Fable 5 has sparked significant controversy following a directive from US Commerce Secretary Howard Lutnick, mandating the suspension of access for foreign nationals. This decision stems from concerns regarding dual-use AI technologies and the potential for misuse in cybersecurity.24
The company, unable to enforce the restrictions effectively, opted to take the model offline entirely. Mythos, a large language model developed by Anthropic, was initially designed for a consortium to enhance cybersecurity but has raised alarms due to its capabilities in identifying vulnerabilities.5
The government’s intervention was prompted by reports of a reliable “jailbreak” that could bypass safety measures, leading to fears of misuse. This incident reflects a broader trend, as the US Bureau of Industry and Security has been tightening controls on advanced AI technologies since 2022, citing military applications in China as a primary concern.69
The Fable episode may signal a shift in how governments will regulate increasingly powerful AI models, indicating that such restrictions could become more common as the technology evolves. As Anthropic navigates these challenges, the implications for AI development and international collaboration remain uncertain.11
“Anthropic's Claude Fable 5 was launched on June 9, but faced immediate restrictions from the US government. The situation underscores the growing scrutiny of dual-use AI technologies and their implications for cybersecurity.”
