AIFortuneabout 2 hours ago
AI models show China censorship bias in study
5 min read
A peer-reviewed Nature study found GPT-4o and Claude Opus gave more Beijing-friendly answers in Chinese than English, with rates up to 88.2%. Meta's Oversight Board found models refused political criticism requests in restrictive countries at 34% versus 14% in freer ones.
Level
Hype check
Tap to vote and see what everyone thinks.
