Acceptable Use Policy

This is Anthropic's Acceptable Use Policy (AUP), version-dated 2026-04-08, governing permissible use of all Anthropic products and services. It applies to all users, including those accessing via authorized resellers or passthrough arrangements. The policy is organized into three tiers: Universal Usage Standards (all users), High-Risk Use Case Requirements (elevated-harm domains), and Additional Use Case Guidelines (chatbots, minors, agentic use, MCP servers).

Not disclosed in this document. This is a usage policy, not a model card, and contains no benchmark scores, modalities, parameter counts, or context window specifications.

Not disclosed in this document.

Not disclosed in this document. No red-team scope, catastrophic-risk evaluations, or threshold findings are described; this is a policy document, not a safety evaluation report.

Anthropic's Safeguards Team implements detection and monitoring to enforce the policy. Violations may result in throttling, suspension, or termination of access; model outputs may be blocked or modified when inputs violate the policy. High-risk use cases require a qualified professional to review outputs before dissemination and mandate disclosure to end-users that AI was involved in producing the content, at minimum at the start of each session. Consumer-facing chatbots must separately disclose AI involvement at the start of each chat session.

The policy covers all users of Anthropic products and services, including access via authorized resellers and passthrough arrangements. Geographic access is governed by Anthropic's Supported Regions Policy. Anthropic may enter contracts with governmental customers that tailor use restrictions to a customer's public mission and legal authorities, provided Anthropic judges that "the contractual use restrictions and applicable safeguards are adequate to mitigate the potential harms addressed by this Usage Policy."

The document acknowledges the policy will be updated "as our technology and the associated risks evolve or as we learn about unanticipated risks," indicating explicit awareness of gaps in current coverage. No technical model limitations are discussed, as this is a policy document.

The document carries a version date of 2026-04-08. No changelog entries, version deltas, or descriptions of changes from a prior version are included in the source text.