The Bottom Line: Claude Fable 5 has been restored with revised security safeguards and is available until July 7 for paid users, with elevated false positive rates in the initial phase.
Anthropic has reactivated the Claude Fable 5 model following discussions with the US government, revising its cybersecurity safeguards in the process. Most coding work is unaffected; new classifiers will initially result in more false positives, but are being continuously refined.
Anthropic has brought back Claude Fable 5 with updated security safeguards. Following consultation with US authorities, the protective measures have been revised. The newly introduced safeguards will, during the transition phase, slightly elevate the number of harmless requests classified as problematic compared to the original Fable safeguards. For flagged requests, users will instead receive a response from Claude Opus 4.8.
The biology and chemistry classifiers remain unchanged. These continue to be calibrated too broadly and trigger fallbacks to Opus 4.8 already on basic biology-related questions. Improvements to these classifiers are already underway. Users receive clear notifications when a request is flagged and can provide feedback via the platforms: in Claude Code using the /feedback command, and on Claude.ai and Cowork via the rating buttons.
Fable 5 is available for all paid tiers with usage quotas through July 7. Users can use the model up to 50 percent of their weekly limit, then switch to other models. Beyond that, Fable 5 can be used via usage credits. This feedback contributes to further optimization of the classifiers and reduction of false alarms.
Source: www.reddit.com · Published July 1, 2026
Lumi AI News — AI-assisted curation in accordance with Article 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.7.3.