Anthropic adds Fable 5.1 to Claude for Government, plans Mythos and Opus for classified use

Anthropic added Fable 5.1 to its Claude for Government platform, positioning it for FedRAMP High compliance, and announced upcoming general availability for Claude for Government Desktop — including Code and Cowork — plus plans to deploy Mythos and Opus models in classified environments. The announcement was made at the Billington Cybersecurity Summit and reflects Anthropic's aggressive public-sector push.

The deal mechanics matter: FedRAMP High and classified-environment support are the gating requirements for serious federal and defense workloads, and Fable 5.1 landing on the government platform (also mirrored onto AWS Bedrock's government-adjacent offerings) signals Anthropic wants to own the compliant-AI niche. Fedscoop reported the news tied to the summit and to Anthropic's broader positioning around AI for cybersecurity.

The context that complicates the story — kept as analysis here, not annexed as a separate event — is that this government expansion runs concurrently with Anthropic's fourth cybersecurity-incident disclosure and reports that it withheld Mythos 5.1 from UK safety testing, plus reported legal tensions with the Pentagon. Selling classified-environment AI while disclosing repeated model-security incidents is a tension critics are quick to note. Watch whether the classified Mythos/Opus deployments actually ship, how the Pentagon relationship evolves, and whether the safety-transparency questions slow procurement.