IT Home On October 9, tech media outlet The Verge published a post today (October 9) stating that Anthropic has revised Claude’s policies and will take effect on November 12, 2026. The first time, a clause prohibiting “continuous and unnecessary abuse or cruel behavior” was added. This rule applies to cases where users repeatedly and without a clear purpose engage in cruel behavior toward the model.
Regarding the enforcement mechanism, Anthropic stated that the main method is to terminate the conversation, but it did not specify when further measures such as user bans will be introduced. It also emphasized that this clause does not apply to common complaints, doubts, dark creative themes, and model testing and research. IT Home has attached the relevant screenshots as follows:
Anthropic has expanded the arms ban to include software and components that enable weapon operation, as well as armed drones and other autonomous vehicles. The company says it has identified multiple attempts to use Claude to develop weapon guidance or control software.
The monitoring provisions explicitly prohibit tracking individuals without consent, whether through real-time tracking or analyzing existing data. Claude shall not be used to decide or suggest targets for investigation, arrest, or prosecution in law enforcement and criminal justice procedures, nor shall it be used to develop or improve monitoring tools.
Anthropic also stipulates that when AI is connected to autonomous physical hardware that could cause harm, a qualified operator must monitor the device and stop it if necessary. For specific government clients, the company can adjust the rules when contractual limitations and safety measures are in place.