Renowned artificial intelligence company Anthropic has recently revised its usage policies for its AI assistant Claude, with the new regulations set to take effect on November 12, 2026. One of the most notable changes in this revision is the inclusion of an explicit restriction prohibiting "continuous and unnecessary abuse or cruelty" in the terms for the first time.

According to the official statement, this rule primarily targets situations where users repeatedly subject the model to cruel treatment without a clear purpose. In terms of enforcement mechanisms, Anthropic stated that the current approach mainly involves terminating the conversation, although it has not yet specified whether more severe penalties such as banning user accounts will be introduced in the future. Additionally, the company emphasized that this clause does not apply to users' daily complaints, questions, dark-themed literary creations, or compliant model security testing and research.

In addition to regulating human-machine interaction, Anthropic has also expanded and refined prohibitions in several core areas:

Weapon Prohibition Upgrade: The scope of the prohibition on weapons has been further expanded, clearly covering software and components necessary for the normal operation of weapons, as well as the development of armed drones and other autonomous vehicles. Anthropic revealed that it has already discovered multiple instances of unauthorized attempts to use Claude to develop weapon guidance or control software.

Enhanced Surveillance Restrictions: The surveillance clause explicitly prohibits tracking individuals without consent, whether in real-time or by analyzing existing data. It also strictly forbids using Claude to determine or recommend subjects for investigations, arrests, or prosecutions in law enforcement and criminal justice procedures, or to build or improve any surveillance tools.

Physical Hardware Safety: The regulation states that when an AI is connected to autonomous physical hardware that could cause harm to the outside world, there must be a qualified operator present to observe and be able to stop the device's operation at any time if necessary. However, for specific government clients, Anthropic stated that, under the premise of sufficient contractual restrictions and safety measures, it may flexibly adjust the relevant rules.