Anthropic Brings Claude to Government: Inside the FedRAMP High Beta

Anthropic has expanded Claude for Government with a public beta of Claude Code and Claude Cowork inside a FedRAMP High authorised environment — the same application commercial customers use, now built specifically for federal agency security and compliance requirements.

What Is Actually Available Now

Claude Code lets public sector technical teams build and modernise the software systems underpinning government services. Claude Cowork works directly with files on the desktop, letting agency staff delegate memo drafting, RFP reviews, casework processing, and deck creation to Claude. Both are delivered through the FedRAMP High authorisation level — the tier required for the most sensitive unclassified federal government data.

The Governance Layer Is the Actual Product

For a government deployment, the AI capability itself is often the easy part — the governance and audit requirements are what actually determine whether an agency can adopt it. Anthropic's expanded offering includes administrator controls to set configuration defaults and control spending allocation across departments, plus tamper-evident audit logs and documentation built to support the formal Authority to Operate (ATO) process security teams require before any system goes live in a federal environment.

This Follows Real State and Federal Deals Already in Place

• California Governor Gavin Newsom announced a partnership bringing Anthropic tools to state agencies to improve services for residents
• The U.S. General Services Administration struck a OneGov deal offering Claude AI access to all three branches of federal government for a nominal $1
• Anthropic has also signed a memorandum of understanding with the UK government
• Anthropic remains the contracted and billing party throughout — agencies do not need a separate cloud provider relationship to get started

What This Signals About Enterprise AI Governance More Broadly

Government procurement is one of the most demanding compliance environments there is — FedRAMP High authorisation, ATO documentation, and tamper-evident audit logging are a considerably higher bar than most commercial enterprise deployments require. Anthropic building this out in public beta, rather than as a bespoke one-off contract, suggests the underlying governance tooling (spend controls, audit logging, configuration policy) is becoming a standard product layer — which is exactly the kind of AI governance infrastructure commercial enterprises in regulated industries should expect to need, and increasingly can adopt off the shelf rather than build themselves.