Anthropic adds abuse protections for Claude as research questions rudeness
Anthropic updated Claude’s usage policy on October 8 to prohibit sustained and unnecessary abuse of AI, according to 36Kr. The company says Claude may end a conversation when a user treats it as an outlet for abuse, while ordinary anger, questioning and model testing remain allowed. The accompanying review cites multiple studies in which politeness, rudeness and threats produced inconsistent results across models and tasks. It concludes that abusive prompting is not a reliable way to improve AI performance.
The policy establishes a reported limit on how users may interact with Claude, while the research summarized by 36Kr and Guokr cautions against treating rudeness as a dependable prompting technique.
最近，AI 公司 Anthropic 干了一件“护犊子”的事儿：以后，不许随便欺负我们家 AI 了。
10 月 8 日，Anthropic 更新了 Claude 的使用政策，专门增加了一条规定，禁止用户“持续且无必要地辱骂或残酷对待”AI。要是有人把 Claude 当成出气筒，它甚至可以主动结束对话。当然，正常的生气、质疑和模型测试都不算违规。
挺招笑的是，Anthropic 自己都不确定，AI 到底有没有“感受”？可能以后等 AI 哪天真觉醒了，Anthropic 能第一时间说：“当年我还哄过你呢！”
还记得刚用 ChatGPT 那会儿，提问时都客客气气地，请问开头，谢谢结束。
图源：aclanthology
#Anthropic#Claude#GPT-4o#GPT-5 nano#Gemini 2.5 Flash#Gemini 2.5 Flash Lite#Claude Opus 4.6#Qwen 3.5