2026 Usage Policy update
Each year, Anthropic updates its Usage Policy in response to the evolving capabilities of our models, and the feedback we’ve received from our customers. We're publishing a new version of the policy today. In this post, we summarize the changes we’ve made.
TL;DR
- New section on deceptive campaigns covers political and commercial deceptive activities, including fake accounts and influence operations.
- Elections section refocused on preventing voter deception and election disruption, renamed 'Do Not Undermine Democratic Processes'.
- Weapons prohibitions now explicitly include software and components for weapons, such as guidance and control systems.
- Surveillance and law enforcement section refined to specify prohibitions on tracking without consent and AI-driven recommendations for investigations.
- High-risk use case requirements for health, finance, and legal matters remain, with added controls for AI connected to autonomous physical actions.
- New prohibition on sustained abusive or cruel behavior towards AI models, applicable in extreme cases.
- Supported Regions Policy updated to clarify enforcement regarding majority ownership and control by entities in unsupported regions.