Anthropic brings Claude Code and Cowork to government desktop

Claude Code and Claude Cowork are available in public beta through Claude for Government Desktop, according to Anthropic.[1] The company says the government desktop product uses the same application as its commercial customers, delivered through a FedRAMP High authorized environment.[1] Anthropic describes Claude Code as a tool for public-sector teams building and modernizing software systems that underpin public services.[1] Cowork works directly with desktop files, which Anthropic says lets agency staff delegate memos, RFP reviews, casework and presentations to Claude.[1] Agencies can request access as new customers.[1]

Anthropic says conversation history is stored locally on the agency-managed device, while inference runs inside a FedRAMP High authorized environment.[1] The application deploys through standard agency mobile device management platforms.[1] Anthropic is publishing a FedRAMP Secure Configuration Guide as a public-facing security resource; its change notification and penetration-test summary are available through its trust center under NDA.[1] Usage exports contain metering data only, which Anthropic says allows agencies to respond to ATO and inspector general requests without moving sensitive material.[1] Sensitive operations on Anthropic’s side require two-person approval, the company says.[1]

Anthropic says usage is purchased in fixed increments with a hard not-to-exceed cap.[1] Administrators can track usage by user and model in the admin console, and automatic burndown alerts warn before a balance runs low.[1] Program offices can connect AI spending to appropriated funds using standard seats or custom seat tiers with spending and model limits.[1] Administrators can set configuration defaults and allocate and control spending across departments.[1] Department-level administrators can allocate seats and prepaid usage to sub-agencies, while each sub-agency manages its own users.[1]

Administrators can use SCIM group mappings to set rate limits, dollar caps and permitted models for specific seat tiers.[1] Layered configuration can also set sub-agency defaults for Claude connections, available features and user-facing instructions.[1] Anthropic says every administrative action is recorded in a hash-chained audit log that organization administrators can review in the product.[1] The company says agencies will receive new capabilities on the same release cadence as commercial users.[1] Anthropic remains the contracted and billing party, so agencies do not need a separate cloud-provider relationship to get started.[1]