Anthropic restored global access to Claude Fable 5 on July 1, 2026, ending a 19-day suspension triggered by a US government export-control order. But the creative writing model that matters most to companion and roleplay users returned with the strongest safeguards Anthropic has ever applied — blocking all NSFW content, erotic roleplay, and mature creative writing outright.
The model was pulled on June 12 after Amazon researchers reported a jailbreak that bypassed Fable 5's cybersecurity safeguards, allowing it to identify software vulnerabilities. The US Commerce Department issued export controls on June 12, cutting off global access. The controls were lifted on June 30 after Anthropic demonstrated a new safety classifier that blocks the reported technique in over 99% of cases.
What Changed With the Restoration
Anthropic's official redeployment post confirms the new classifier is the primary change. When the classifier detects a blocked request, it redirects to Claude Opus 4.8 instead of refusing outright — a design choice meant to reduce user friction. But the company acknowledged the classifier may increase false positives on legitimate coding and debugging tasks.
For users who care about creative writing and companion-style interactions, the restoration changes nothing about the model's content restrictions. Fable 5 was launched with "the strongest safeguards we've ever applied to a model," Anthropic stated in its redeployment announcement. Direct testing confirms the model refuses all sexually explicit creative writing, intimate roleplay, and romantic scene generation — regardless of how prompts are framed.
Anthropic's Claude Mythos 5, the underlying model without safety classifiers removed, remains restricted to approved US organizations through Project Glasswing — a collaboration with the US government for defensive cybersecurity. It is not available to the public or to developers building companion applications.
Access and Pricing
Fable 5 is available on Claude.ai, the Claude Platform, Claude Code, and Claude Cowork. For Pro ($20/month), Max, Team, and select Enterprise plans, Fable 5 usage was capped at 50% of weekly limits through July 7, after which it shifted to usage credits. Anthropic is also restoring access via AWS, Google Cloud, and Microsoft Foundry.
The model supports a 1 million token context window and 128,000 maximum output tokens — making it one of the most capable chat models available for long-form creative work. But those capabilities operate within hard content boundaries.
What This Means for Companion AI
The Fable 5 episode is a case study in the widening gap between what AI models can do and what their platforms will let them do. The suspension was triggered by a cybersecurity jailbreak, not NSFW content, but the outcome is the same: Anthropic has signaled it will pull models offline and add layers of restriction when pressured by the government.
For companion AI users and developers building on Anthropic's platform, the message is clear. The company's constitutional AI approach and its willingness to cooperate with export controls mean that unrestricted creative expression — erotic roleplay, mature storytelling, uncensored character interactions — will not find a home on Anthropic's infrastructure. The Mythos 5 model, which would provide the uncensored experience, is locked behind government vetting and not coming to the public market.
This stands in contrast to the broader companion industry, where platforms like Character.AI, Replika, and open-source models have carved out space for adult-oriented interactions. Anthropic's decision to keep Fable 5 locked down, even after restoration, reinforces the divide between enterprise-safe AI and the tools that companion users actually want.