Anthropic Says It’s Taking Claude Fable 5 Offline to Comply With US Government Order

Details

Anthropic received a letter at 5:21pm ET Friday from the US government directing it to suspend access to Claude Fable 5 and Mythos 5 for 'any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.' Rather than attempt selective enforcement, Anthropic pulled access for all customers globally.\n\nThe government's stated concern is a jailbreak that could enable the models' cybersecurity capabilities to be misused. Anthropic disputes the severity: the company says it reviewed a demonstration and found it narrow and non-universal, capable of surfacing only 'previously known, minor vulnerabilities' that other public models can also find without any bypass.\n\nThis is the latest escalation between Anthropic and the Trump administration. Earlier in 2026, the DoD labeled Anthropic a 'supply chain risk,' barring government contractors from using Claude — prompting Anthropic to sue. Fable 5 was the company's first publicly available Mythos-class model, built specifically with safeguards around cybersecurity, biology, and chemistry to enable public access to previously restricted capabilities.\n\nThe incident raises urgent questions about the government's ability to unilaterally suspend commercial AI products and whether export control frameworks are equipped to handle frontier AI models.