Study Finds Pro-Beijing Bias in Popular Chinese AI

yahoo.com —

A study by Israeli cybersecurity startup Hirundo found that Alibaba's Qwen, the world's most popular free AI model, shows pro-Beijing bias or censorship in 89.8% of politically sensitive responses, according to CBS News. The model, downloaded over 3 billion times and used by Uber and Airbnb, denied forced labor camps for Uyghurs and refused to confirm force used in Tiananmen or Hong Kong protests. It also labeled Falun Gong a "dangerous cult." Chinese open-weight models grew from under 2% to over 45% of global usage by June. Hirundo created a "Westernized" Qwen version, cutting biased responses to 2.8%, while Booz Allen recommended banning such models over security vulnerabilities.


With a significance score of 4.3, this news ranks in the top 3.7% of today's 29638 analyzed articles.

Get summaries of news with significance over 5.5 (usually ~10 stories per week). Read by 10,000+ subscribers:


Study Finds Pro-Beijing Bias in Popular Chinese AI | News Minimalist