The biggest changes Anthropic just made to Claude’s rules

Anthropic’s new rule against abusing Claude wasn’t the only update to its revised usage policy for the AI chatbot.

The artificial intelligence lab has also spelled out a series of new usage provisions about deceptive campaigns, elections, weapons development, surveillance, law enforcement, and Claude-connected hardware.

Anthropic said this week that most of the changes in the latest version of its usage policy were intended to clarify existing rules.

The updated policy takes effect on November 12.

Anthropic added a new section to its AI rules titled, “Do Not Engage in Deceptive Campaigns or Artificial Activity,” which the company said applies to deceptive activity of any kind, political or commercial. “It covers efforts to obscure who is behind a message or amplify content through fake accounts or posts, along with building the tools and infrastructure for running influence operation campaigns,” Anthropic said.

The company refined its political content restrictions and sharpened its elections section, focusing it on “disallowing the use of Claude to deceive voters or disrupt elections.” That section was renamed “Do Not Undermine Democratic Processes” and includes using Anthropic’s products to “generate or disseminate false or misleading information in political and electoral contexts.” Anthropic said its usage policy has always barred the use of Claude to make weapons.

However, its new rules make explicit that the chatbot cannot be used to develop the software or components that make weapons work, “as well as actions like arming drones or other autonomous vehicles.” The company said it has recently seen users attempt to use its models to “build guidance and control software for weapons.” Anthropic tightened its usage policy’s surveillance and criminal justice section to more specifically lay out banned activity.

That includes tracking people without their consent as well as doxxing.

Under the new policy, Claude also cannot be used to “decide or recommend who to investigate, arrest, or charge in a law enforcement or criminal justice process,” or to build or improve tools designed for surveillance.

Anthropic updated its policy to reflect new requirements for its models when they’re connected to hardware that “takes autonomous physical actions and which might be capable of causing injury.” “A qualified operator must be able to observe the equipment and stop it if needed; the equipment must also be able to hold a safe state if Claude is disconnected,” Anthropic said.