Statement on the US government directive to suspend access to Fable 5 and Mythos 5
Short summary
The US government issued an export control order to cut off all foreign-national access to Anthropic’s Fable 5 and Mythos 5, citing a potential jailbreak vulnerability. Anthropic says the reported exploit is narrow, already known across other models, and disagrees that it warrants a full suspension. The company plans to share more details within 24 hours and restore access if possible.
- The US government ordered Anthropic to cut off all foreign-national access to Fable 5 and Mythos 5 over a claimed jailbreak vulnerability, with only a "national security concerns" explanation given.
- Anthropic says the exploit is just a narrow codebase-scanning jailbreak trick that only surfaces minor, already-known vulnerabilities present in any public model, including OpenAI's GPT-5.5.
- Anthropic disabled both models entirely despite disputing the severity, arguing the shutdown is disproportionate and could chill AI releases industry-wide.
- The company is pushing for a transparent process with clear technical criteria instead of sudden verbal directives, and plans to release more details within 24 hours while seeking to restore access.
Full summary
The US government just issued an export control order forcing Anthropic to cut off all foreign‐national access to its Fable 5 and Mythos 5 models, inside or outside the U.S. Anthropic got the directive at 5:21 pm ET and had no detailed explanation beyond a reference to “national security concerns.” In response, the company immediately disabled those two models for everyone, even though access to its other AI systems remains intact.
Anthropic believes the government’s concern centers on a single jailbreaking method that asks Fable 5 to scan and repair a codebase. The company reviewed that technique and found it only exposed minor, already-known vulnerabilities—ones any publicly available model can produce. Anthropic ran extensive red-team tests before launch, worked with U.S. and U.K. agencies plus private partners for thousands of hours, and never saw a universal jailbreak that broadly bypassed safeguards. Their defense-in-depth setup aims to make attacks narrow or costly, backed by 30-day data retention for rapid detection and response.
Even so, Anthropic views the government’s move as disproportionate. They argue that pulling a commercial model over a narrow exploit, one they’ve shown is widespread—including in OpenAI’s GPT-5.5—would stifle new AI releases across the industry. The company wants a formal, transparent process centered on clear technical criteria rather than a sudden verbal order. They plan to share more technical details within 24 hours and hope to restore Fable 5 and Mythos 5 access soon.