AI News · Hot news ·
Anthropic updates usage policy on deception, surveillance and autonomous hardware
Anthropic published its 2026 Usage Policy update, which takes effect on November 12. It consolidates rules on deceptive campaigns, refocuses its elections section, and clarifies weapons and surveillance prohibitions. It adds requirements for Claude connected to hardware that takes autonomous physical actions. A new rule bans sustained and needless abusive or cruel behavior toward Claude models.
Key points
- Anthropic's updated Usage Policy takes effect on November 12, 2026.
- Deceptive campaign rules are consolidated into one section.
- Claude cannot be used to decide or recommend who to investigate, arrest or charge.
- New safety requirements apply to Claude controlling hardware that acts autonomously.
- A new rule bans sustained and needless abusive behavior toward Claude models.
What happened: Anthropic published its 2026 Usage Policy update on October 8. The new policy takes effect on November 12. Anthropic said it updates the policy each year in response to model capabilities and customer feedback, and that "Most of the updates in the latest version are intended to clarify existing rules."
The details: Rules on fake accounts, fabricated news sites and similar activity, which were previously spread across several sections, are now consolidated under "Do Not Engage in Deceptive Campaigns or Artificial Activity." The section covers political and commercial deception, including hiding who is behind a message and building tools for influence campaigns. The elections section is renamed "Do Not Undermine Democratic Processes" and focuses on deceiving voters or disrupting elections. Anthropic removed a blanket ban on personalized vote and campaign targeting to protect legitimate civic work, while deceptive targeting and misuse of voter data remain prohibited.
The weapons rules now explicitly cover software and components that make weapons work, and actions such as arming drones. The surveillance section prohibits tracking people without consent, whether in real time or through analysis of past data. Claude cannot be used to decide or recommend who to investigate, arrest or charge. Permitted uses include consented tracking such as fraud monitoring, plus content moderation, journalism and legal research. Anthropic says this is a clarification, not a change in enforcement.
The human-in-the-loop and disclosure requirements for high-risk uses are unchanged, but the policy now specifies which recommendations are covered. New requirements apply when Claude is connected to hardware that takes autonomous physical actions and could cause injury: a qualified operator must be able to observe and stop the equipment, and the equipment must hold a safe state if Claude is disconnected.
A new prohibition covers "sustained and needless abusive or cruel behavior" toward Claude models, Engadget reported. Anthropic says it applies only to extreme cases with no discernible purpose and excludes ordinary frustration, pushback, dark creative themes and model testing. Ending persistently abusive conversations remains the main enforcement tool. The update also clarifies that entities majority-owned or controlled from unsupported regions cannot use Claude.
What to watch: Companies building products on Claude, particularly in public safety, marketing, robotics or political communication, should check their use cases against the new text before the November 12 effective date.
Our take
Any team building on Claude should review the new rules before November 12, especially for law enforcement, tracking, marketing outreach or robotics use cases.