Anthropic has restored global access to its Claude Fable 5 model, one day after the U.S. Department of Commerce withdrew the export controls it had placed on the system on June 12. The resolution came after Anthropic deployed a single safety filter targeting one specific prompting technique, with the Commerce Department’s Center for AI Standards and Innovation reviewing the safeguards before the controls were lifted. Fable 5 is available again across Claude.ai, the Claude Platform, Claude Code, and Claude Cowork, with access through AWS, Google Cloud, and Microsoft Foundry set to follow.<\/p>\n
The June 12 directive barred any foreign national, including Anthropic’s own non-citizen staff, from using Fable 5 or the more capable Mythos 5 model that Fable 5 is built on. Because Anthropic had no way to verify users’ nationalities, it pulled both models globally rather than attempt geographic compliance.<\/p>\n
The specific concern originated with Amazon researchers, who reported a way to prompt Fable 5 into identifying software vulnerabilities and, in one instance, writing code that demonstrated how a vulnerability could be exploited.<\/p>\n
Anthropic trained a new classifier that blocks the reported technique in more than 99 percent of cases and reroutes flagged requests to the older Opus 4.8 model. The company acknowledged that the classifier also catches some benign coding and debugging requests as a side effect.<\/p>\n
The filter targets the prompt pattern rather than the model’s underlying capabilities. Fable 5 can still identify the vulnerabilities described in the Amazon report; the system detects the request and reroutes it instead of stripping the ability from the model itself.<\/p>\n
This approach has a known weakness. Detection-based safeguards are what were defeated to trigger the export controls in the first place, and a classifier tuned to one technique does not address techniques that have not yet been discovered. Anthropic has stated that no model can be made fully robust to jailbreaks and that it expects more to surface over time.<\/p>\n
Commerce withdrew the controls after joint testing with Anthropic and Amazon found that the vulnerabilities in question were not unique to Fable 5. The review concluded that Opus 4.8, OpenAI’s GPT-5.5, and China’s Kimi K2.7 could identify the same vulnerabilities. Every model tested, including Haiku 4.5, Sonnet 4.6, and several Opus versions, was able to reproduce the single exploit demonstration, supporting the position that so-called Mythos-class cyber capabilities had been overstated.<\/p>\n
Fable 5’s return reclaims benchmark positions that Chinese lab Z.ai’s GLM-5.2 had held by default during the 18-day outage, including the top accessible score on the AA-Briefcase multi-week task test.<\/p>\n
Mythos 5, which carries fewer guardrails and remains limited to Project Glasswing partners, returned to a set of U.S. organizations on June 26.<\/p>\n
Anthropic has opened a HackerOne program for researchers to report new Fable 5 jailbreaks. The company has also committed to giving designated government partners earlier access to test future frontier models before public release.<\/p>\n
For Pro, Max, Team, and select Enterprise plans, Fable 5 counts toward up to 50 percent of weekly usage limits through July 7. After that date, it moves to usage credits.<\/p>\n
A June 12 U.S. Department of Commerce directive barred any foreign national, including Anthropic’s own non-citizen staff, from using Fable 5 or Mythos 5. Because Anthropic could not verify users’ nationalities, it pulled both models globally.<\/p>\n
Anthropic deployed a safety filter that blocks the specific prompting technique in more than 99 percent of cases and reroutes flagged requests to Opus 4.8. The Commerce Department’s Center for AI Standards and Innovation reviewed the safeguards before the controls were withdrawn.<\/p>\n
Fable 5 is back on Claude.ai, the Claude Platform, Claude Code, and Claude Cowork, with access through AWS, Google Cloud, and Microsoft Foundry set to follow.<\/p>\n