Anthropic bans using Claude to track people without consent

Companies and individuals who use Claude will not be able to use it to track someone without their consent starting Nov. 12. That is what Anthropic's usage policy update, published Oct. 8, establishes. The policy is the document that sets what can be done with the model.

The section on surveillance and law enforcement was rewritten. The text bars real-time tracking and analysis of previously collected data, and building or improving surveillance tools. In criminal justice the rule is explicit: Claude cannot decide or recommend whom to investigate, arrest or charge. Uses with consent, such as fraud monitoring, remain allowed, along with content moderation, journalism and legal research.

The update touches other topics. A new section covers deceptive activity, including fake accounts, fabricated news sites and influence operations. The elections section is now titled "Do Not Undermine Democratic Processes" and focuses on deceiving voters or disrupting elections; Anthropic dropped the blanket ban on personalizing campaign messages, and misuse of voter data remains barred by the privacy and surveillance sections.

For hardware that acts on its own in the physical world, a qualified operator must be able to observe and stop it, and the equipment must stay in a safe state if Claude disconnects. The policy also adds a ban on sustained, unnecessary cruelty toward the models, which does not cover criticism or testing.

The policy is a company rule: it is not a law and no authority supervises it. This report relied only on Anthropic's text, with no third-party coverage and no line-by-line comparison with the previous version. The announcement does not explain how the rules will be enforced or whether there will be exceptions for government contracts.

Disclosure: part of El Mediático's operation runs on Anthropic models.