The AI company Anthropic has published an update to its Usage Policy on its flagship AI model Claude. The new update will take effect on November 12, 2026, consolidating and clarifying rules surrounding influence operations, civic engagement, defense, and surveillance.
The most prominent structural addition is a dedicated section titled “Do Not Engage in Deceptive Campaigns or Artificial Activity”, which brings together previously scattered restrictions regarding fake accounts, fabricated news sites, and coordinated influence infrastructure across both political and commercial contexts. Alongside this, Anthropic renamed its elections framework into “Do Not Undermine Democratic Processes”, zeroing in on voter deception and turnout disruption while lifting a prior blanket ban on personalized outreach to allow legitimate civic uses like multilingual voter information and ballot notices.
The company also announced that it expanded the definitions in its weapons and surveillance policies. The weapons prohibition explicitly covers guidance software, critical functional components, and the arming of drones or autonomous systems. In surveillance and criminal justice, the policy formally prohibits using models to track individuals without consent, automate criminal investigation or charging decisions, or develop dedicated surveillance tools, while keeping carve-outs (the partial separation, spin-off, or sale of a business unit, subsidiary, or specific asset from a larger parent company) for authorized activities such as fraud monitoring, content moderation, and journalism.
