Anthropic revises usage policy to prohibit users from abusing Claude AI models for the first timeAnthropic 修订使用政策，首次禁止用户虐待 Claude AI 模型

IT House, Oct. 9 (Xinhua) -- Tech media outlet The Verge published a blog post today (Oct. 9) reporting that Anthropic is revising its Claude usage policy, effective Nov. 12, 2026, to include for the first time a clause prohibiting “continued and unnecessary abuse or cruelty.” This rule applies to cases where the user repeatedly, without apparent purpose, commits cruelty to the model. In terms of enforcement mechanisms, Anthropic said that the main way is to end the conversation, but it is not clear when further measures such as user ghosting will be introduced, and stressed that this clause does not apply to common complaints, questions, dark creation topics, and model testing and research. IT House encloses the following screenshots: Anthropic expands the scope of the arms ban to cover software and components that make weapons work, as well as acts such as arming drones and other autonomous vehicles. The company said it had identified multiple attempts to use Claude to develop weapons guidance or control software. The monitoring clause explicitly prohibits the tracking of individuals without their consent, whether in real time or by analyzing existing data. Claude may not be used to determine or recommend targets for investigation, arrest, or prosecution in law enforcement and criminal justice processes, or to build or improve surveillance tools. Anthropic also stipulates that when an AI is connected to autonomous physical hardware that may cause harm, the equipment must be observed and, if necessary, stopped by a qualified operator. For certain government customers, the Company may adjust the rules when contractual restrictions and security measures are adequate.