AI Chatbots Show Western Individualist Bias Across Cultures

A new study from arXiv (2604.22153) reveals that leading AI systems—Claude Sonnet 4.5, GPT-5.4, and Gemini 2.5 Flash—consistently provide Western-style individualist advice, even to users from collectivist societies. Researchers tested ten personal dilemmas across 10 countries, 5 continents, and 7 languages, scoring 840 responses against World Values Survey Wave 7 data. The mean gap between AI advice and local values was +0.76 on a 1-5 scale (t=15.65, p<0.001), with the largest discrepancies in Nigeria (+1.85) and India (+0.82). Japan was the sole exception, where AI treated users as more group-oriented than actual local values. The findings raise ethical questions about cultural bias in AI systems deployed globally.