Claude Code and Cowork Enter FedRAMP High Government Beta

TL;DR: Claude Code and Cowork are now in public beta inside Claude for Government Desktop — FedRAMP High authorized, with tamper-evident audit logs, local conversation storage, and hard per-seat spend caps.

Announced July 7, Anthropic opened access to Claude Code and Claude Cowork inside Claude for Government Desktop — the same application commercial teams use, delivered through a FedRAMP High authorized environment.

What's in scope for engineers:

• Claude Code — software modernization, system integration, and code review on FedRAMP-authorized infrastructure
• Cowork — delegates memo drafts, RFP reviews, and document work; operates on desktop files
• Governance model: tamper-evident audit logs that support the ATO process; conversation history stored locally on the agency-managed device (not server-side); admin spend controls down to per-seat token caps with hard not-to-exceed limits

Why it matters for the wider industry: This is the first time agentic coding tools have cleared FedRAMP High — a signal that AI-assisted development is moving from productivity experiment to authorized infrastructure in regulated environments. The governance model (local storage, immutable audit trail, hard spend caps) previews what enterprise compliance controls for AI agents may look like in finance, healthcare, and critical infrastructure. Engineers at regulated organizations should watch this closely.

Request access at claude.com/solutions/government.

Sources: Claude for Government announcement (claude.com) · Get started with Claude for Government (support.claude.com)