Anthropic Expands Claude Availability for Government Agencies

Anthropic has launched Claude 3 Haiku and Claude 3 Sonnet models for use in the U.S. intelligence community (IC) on AWS Marketplace and on AWS GovCloud. This allows government agencies to leverage AI across various areas such as improving citizen services, streamlining document review, data-driven policymaking, disaster response, and public health.

Anthropic has also adjusted its service agreements to meet the government's special requirements. In particular, it has established exceptions to its Usage Policy to support legitimate foreign intelligence analysis, such as combating human trafficking or providing advance warning of military activity. However, activities such as spreading disinformation, weapons design, censorship, and malicious cyber operations remain strictly prohibited.

This policy applies first to models at AI Safety Level 2 (ASL-2) under Anthropic's Responsible Scaling Policy (RSP). Anthropic plans to continue working with governments—including collaborating with the UK AI Safety Institute (UK AISI)—to build testing and measurement frameworks for a safe transition to AI.