Anthropic bans non-consensual tracking tools built on Claude from November 12

Summary

• Anthropic's 2026 usage policy bars building or improving surveillance tools on Claude, effective November 12.
• Claude may no longer recommend who to investigate, arrest or charge, the company said.
• Anthropic also dropped its blanket ban on personalized vote and campaign targeting.

The latest

Building or improving tools designed to track people without their consent will be prohibited on Claude from November 12, under a usage policy update Anthropic announced on October 8, 2026. The company said real-time tracking and analysis of previously collected data both fall under the ban, and that Claude cannot be used to decide or recommend who to investigate, arrest or charge.

Details

• The surveillance rule: Anthropic said Claude may not be used to track individuals without consent, whether in real time or by analyzing data gathered earlier, and may not be used to build or improve tools designed for surveillance. The company said consent-based tracking stays permitted, naming fraud detection, content moderation, journalism and legal research as examples.
• Physical equipment: Where Claude is connected to equipment that takes autonomous physical actions capable of causing injury, Anthropic said a qualified operator must be able to observe the equipment and stop it if needed. The policy also requires such equipment to remain in a safe state if Claude's connection drops.
• Weapons expanded: The updated policy bars using Claude to build the software and components that make weapons function, broadening earlier language. Anthropic added an explicit prohibition on weaponizing drones and other autonomous vehicles.
• Elections narrowed: The elections section was renamed Do Not Undermine Democratic Processes and now targets the use of Claude to deceive voters or disrupt elections. Anthropic listed false information about candidates or voting procedures, impersonation of candidates or election officials, and suppressing turnout as covered examples.
• Targeting unbanned: Anthropic said it removed its blanket prohibition on personalized vote and campaign targeting. Targeting that relies on deception or on misuse of voters' personal data remains prohibited under other sections of the policy.
• Deception section: A new section, Do Not Engage in Deceptive Campaigns or Artificial Activity, consolidates previously scattered rules and covers deceptive activity of any kind, political or commercial. It names concealing who is behind a message, amplifying content through fake accounts or posts, and building tools or infrastructure for influence operations.
• Model abuse: The policy prohibits sustained and needless abusive or cruel behavior toward Anthropic's models. The company said this applies only in extreme cases and excludes ordinary frustration, disagreement, dark creative subject matter, and model testing or research, with Claude's ability to end persistently abusive conversations remaining the main enforcement mechanism.
• Geographic scope: Anthropic clarified that use is barred for people physically located in unsupported regions, for entities registered or headquartered there, and for entities majority-owned or controlled by persons or entities in those regions.
• What is unchanged: Anthropic said several changes, including the rewritten weapons and surveillance clauses and the high-risk use-case requirements, clarify existing rules rather than alter enforcement. Requirements for a qualified human in the loop and for notifying affected individuals in high-risk use cases were left unchanged.

Between the lines

The two headline shifts move in opposite directions. The law-enforcement and surveillance clauses tighten what institutional customers can build, down to barring recommendations on who to arrest. The elections change loosens a rule, replacing a blanket ban on personalized campaign targeting with conduct-based limits that still catch deception and voter-data misuse — a narrower perimeter policed case by case rather than category-wide.

What's next

The updated policy takes effect on November 12, when the surveillance, weapons and physical-equipment clauses become enforceable. Watch how Anthropic applies the consent test to existing law-enforcement and security customers after that date.

Source: Anthropic