Anthropic Updates Claude Usage Rules and Bans Model Abuse

Anthropic has released a comprehensive update to its usage policy, set to take effect on November 12. The revised guidelines address the expanding capabilities of its Claude models, which are increasingly performing longer and more autonomous tasks. For developers and enterprise users, the update clarifies boundaries around high-risk applications, political engagement, and physical hardware integration, while consolidating rules against deceptive campaigns and artificial activity.

The policy refines political restrictions by replacing the blanket ban on personalized campaign targeting with a narrower focus on preventing voter deception under a section renamed Do Not Undermine Democratic Processes. This allows legitimate civic applications, such as translating voter information. Meanwhile, the company explicitly bans using Claude to build guidance or control software for weapons, arm drones, or conduct non-consensual surveillance. Under the new rules, Claude cannot be used to recommend arrests or build surveillance tools, though consensual tracking for fraud monitoring remains permitted.

For physical deployments, the policy aligns with Anthropic's Model Hardware Standard. If Claude is connected to physical machinery capable of causing injury, a qualified operator must be able to halt the system, and the hardware must default to a safe state if disconnected. Additionally, the update introduces a novel prohibition against sustained, cruel behavior toward the AI. Extreme cases of abuse will allow Claude to autonomously terminate conversations on Claude.ai and Claude Code.

Alongside the policy changes, Anthropic announced a $150 million commitment over three years to the Genesis Mission, a federal initiative designed to accelerate scientific discovery through AI. The company is also launching the Anthropic Cyber Mission and expanding its Cyber Verification Program, which provides advanced cyber capabilities and reduced blocking classifiers to eligible security professionals.