From Anthropic · · 1m 30s
Anthropic 2026 Policy Update
Anthropic updates its AI policy to address deceptive campaigns, weapons risks, and surveillance.
In brief
Anthropic updates its AI policy to address deceptive campaigns, weapons risks, and surveillance. Anthropic released a 2026 Usage Policy update clarifying rules for deceptive activity, weapons development, and high-risk use cases. Originally reported by Anthropic.
2026 Usage Policy Update
Anthropic publishes a new Usage Policy update on October 8, 2026.
Clarifying Rules for New Capabilities
The update clarifies rules for Claude’s longer, more independent work and new misuse patterns in influence operations, weapons development, and surveillance.
New Section on Deceptive Campaigns
Anthropic created a new section titled Do Not Engage in Deceptive Campaigns or Artificial Activity to cover fake accounts and fabricated news sites.
Refocused Elections Section
Anthropic renamed its elections section to Do Not Undermine Democratic Processes and removed personalized vote targeting prohibitions.
Clarifying Weapons Prohibitions
The policy now explicitly prohibits using Claude to build guidance software for weapons and autonomously arm drones.
Precise Surveillance Rules
Anthropic prohibits tracking people without consent, using Claude to decide arrests, and building surveillance tools.
High-Risk Use Case Requirements
Anthropic requires a human in the loop for health, finance, and autonomous physical actions to prevent harm.
Abusive Behavior Prohibition
Anthropic added rules against sustained, needless abusive behavior toward Claude models in extreme cases.
Supported Regions Policy Update
Anthropic clarified enforcement of its Supported Regions Policy for companies headquartered in unsupported regions.
Future Policy Updates
Anthropic will continue updating its policy as Claude’s capabilities and risks evolve.
The headline
Anthropic released a 2026 Usage Policy update clarifying rules for deceptive activity, weapons development, and high-risk use cases.
