Anthropic's Claude Fable 5 Was Banned by the US Government Just 3 Days After Release — Here's Why

Anthropic released Claude Fable 5, a Mythos-class model that quickly became the most capable AI ever made generally available. It achieved state-of-the-art results across nearly all tested benchmarks, including software engineering, scientific research, and vision. Users reported it felt like a 'thought and design partner' rather than just a coding agent, with methodical debugging, judgment, and taste. Boris Cherny (creator of Claude Code) called it 'the biggest step up since Opus 4.5' and noted he uninstalled his IDE after a few weeks because Claude handled all coding tasks. The model's 'big model smell' — a sense of dimensionality and trustworthiness — was widely praised.

However, within three days of launch, the United States Government forced Anthropic to take Fable 5 offline after a jailbreak exploit was brought to their attention. This marked a shift from previous instances where experts could jailbreak models but only in laboratory contexts; now, a practical vulnerability triggered immediate government intervention. The model was priced at $10 per million input tokens and $50 per million output tokens (double Claude Opus), required a 30-day data retention policy, and could be accessed via Claude Code or API under the name claude-fable-5. Anthropic is expected to re-release the model within two weeks after addressing the jailbreak. The incident highlights growing regulatory oversight of frontier AI capabilities.