Anthropic's Fable 5 faces US export ban after alleged jailbreak by foreign nationals
US government orders immediate suspension of Fable 5 and Mythos 5 access worldwide.
Anthropic's latest model, Fable 5, was released two months after the company warned about its predecessor, Mythos 5, being too dangerous for public use due to advanced cybersecurity capabilities. Fable 5 included guardrails designed to prevent misuse. However, the US government has now ordered Anthropic to immediately disable access to both Fable 5 and Mythos 5 for all foreign nationals — including Anthropic's own foreign employees — citing national security concerns over a discovered jailbreak method. Anthropic claims the demonstrated technique only revealed a few minor, previously known vulnerabilities that other publicly available models can also find without requiring a bypass. The company argues that non-universal jailbreaks are inevitable and narrow, and that the specific method was reported by Amazon, which is both an investor and inference provider to Anthropic. Senior Anthropic staff are currently in Washington D.C. trying to resolve what they call a misunderstanding, while White House officials accuse the company of insouciance to legitimate security risks. The conflict highlights the growing tension between AI developers promoting safety and governments fearing misuse of frontier models.
- US government ordered Anthropic to suspend Fable 5 and Mythos 5 for all foreign nationals worldwide, including foreign employees.
- The alleged jailbreak, reported by Amazon, only uncovered minor, known vulnerabilities that other public models can also find.
- Anthropic is negotiating in Washington D.C., disputing the government's national security justification.
Why It Matters
This conflict sets a precedent for government control over advanced AI models, impacting global access and innovation.