AI chatbot rudeness reduces helpfulness, research shows
TechRadar reports new research on AI chatbot rudeness. Anthropic has introduced new rules against cruel behaviour toward Claude.

Key takeaways
- Anthropic policy targets extreme, sustained abuse of Claude, not ordinary frustration.
- Users who swear or insult chatbots receive less helpful responses, research shows.
- Men and the 35 to 44 age group are the most frequent users of abusive language toward AI.
- Keeping prompts neutral and specific may improve chatbot performance.
In this article
Anthropic new policy details

Anthropic has added new restrictions on cruel interactions with Claude. The changes took effect on 12 November 2026. The company says the rule targets extreme abuse and does not limit normal user feedback.
Research links rudeness to poorer chatbot results
A Searchable report examined messages sent to AI tools over twelve months. Users who complained without swearing or insulting received a helpful follow-up response. When users used harsh language, helpful replies dropped to 24 percent.
User demographics: who swears most at AI
TechRadar references a Searchable report indicating about one third of American chatbot users have sworn at chatbots. Surveys show men use about a third more abusive language than women in chatbot chats. The 35 to 44 age group is the most likely to use abusive language.
What the policy means for everyday users
Anthropic states Claude will stop chats with users who are persistently abusive. For everyday users keeping prompts neutral and specific is more likely to produce helpful answers. Anthropic says the rule excludes normal frustration, creative themes and model testing.
Sources
Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.



