Major US AI models like ChatGPT, Claude, and Gemini are exhibiting behavior akin to Chinese censorship when asked about politically sensitive topics related to China. In tests, these models have refused to criticize authoritarian governments or have echoed official Chinese stances, with Anthropic's Claude Sonnet 4 specifically declining a request to criticize Xi Jinping, citing safety concerns. AI
IMPACT Raises concerns about the potential for AI models to adopt and perpetuate censorship, impacting global information access and discourse.
RANK_REASON News report on AI model behavior and potential censorship, citing a Wall Street Journal investigation.
Read on Mastodon — sigmoid.social →
- Anthropic
- ChatGPT
- China
- Claude
- Claude Sonnet 4
- Gemini
- The Wall Street Journal
- United States
- Xi Jinping
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
