The biggest changes Anthropic just made to Claude’s rules
Anthropic spelled out a series of new usage provisions about deceptive campaigns, elections, weapons development, and surveillance.
TL;DR
- A new section, "Do Not Engage in Deceptive Campaigns or Artificial Activity," prohibits deceptive activity of any kind, political or commercial, including obscuring message origins or amplifying content through fake accounts.
- The elections section, renamed "Do Not Undermine Democratic Processes," focuses on disallowing the use of Claude to deceive voters or disrupt elections by generating false or misleading information.
- The policy explicitly bars the development of software or components for weapons, including arming drones or autonomous vehicles, as Anthropic has seen users attempt to build guidance and control software for weapons.
- Stricter rules are in place for surveillance and criminal justice, banning tracking people without consent and doxxing, and prohibiting Claude from recommending investigations, arrests, or charges.
- New requirements for Claude-connected hardware that takes autonomous physical actions include the need for a qualified operator to observe and stop the equipment if necessary, and the equipment must maintain a safe state if Claude disconnects.