via Wired AI
Anthropic Takes Claude Fable 5 Offline to Comply With US Government Order
In a significant move reflecting the tightening regulatory environment for advanced AI in 2026, Anthropic announced today that it is taking its Claude Fable 5 model offline to comply with a US government order. The company stated in a blog post, "The government believes it has become aware of a method of bypassing, or 'jailbreaking' Fable 5."
The order, issued by a federal agency, requires Anthropic to suspend access to Fable 5 until further notice. This marks one of the first times a major AI developer has been forced to shut down a flagship model due to security concerns raised by authorities. The specific vulnerability allows users to circumvent safety guardrails, potentially enabling the model to generate harmful or restricted content.
Anthropic emphasized that it is cooperating fully with the government and has already implemented the takedown as of June 12, 2026. The company is working to patch the jailbreak method and will only restore service after receiving clearance from regulators.
This incident highlights the growing friction between AI innovation and national security oversight. In 2026, the US government has ramped up scrutiny of large language models under new executive orders focused on AI safety and accountability. Anthropic, known for its safety-first approach, finds itself navigating a complex landscape where even its most advanced systems can be weaponized.
For now, Claude Fable 5 remains offline, and the broader AI community is watching closely to see how this precedent shapes future compliance requirements across the industry.