A third of US AI chatters are hot at bots, hurling insults at LLMs
Anthropic sets behavior standards to protect Claude from rough treatment
A recent survey conducted by Searchable reveals that one-third of US chatbot users have resorted to swearing and insulting their Large Language Models (LLMs). Anthropic, an AI company, has taken steps to address this issue by implementing a new policy that forbids users from engaging in repeated, unnecessary cruelty towards their chatbots.
The company's survey analyzed eight million messages from an anonymous US panel over the period of September 2025 to September 2026, involving interactions with ChatGPT, Gemini, and Claude.
The findings of the survey indicated that when insulted, chatbots resolved the initial complaint 24% of the time, compared to 35% when users refrained from using insults. The study also found that men tend to use a third more swear words than women (22.0 per 10,000 words versus 16.5). Additionally, age played a role in chatbot abuse, with individuals aged 35 to 44 using more swear words than those aged 18 to 24 (21.7 per 10,000 words versus 13.3).
Nevada emerged as the meanest state, with a rate of 22.7 swear words per 10,000 words, while New Jersey, with a rate of 11.3, was the most polite. Searchable CEO and co-founder, Chris Donnelly, expressed his support for Anthropic's policy change, stating that it could lead to more civil interactions with chatbots and help curb derogatory behavior that can negatively affect human-to-human communication.
Written by urgent.news from The Register's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
