Chinese AI systems censor Tiananmen Square history
Chinese AI chatbots including DeepSeek, Qwen3, and Doubao reportedly refuse to provide information about the 1989 Tiananmen Square military crackdown, blocking search results or returning error messages about the historical event.
What happened
Chinese AI chatbots including Qwen3 from Alibaba, Doubao from Bytedance, and DeepSeek block or refuse to provide information about the 1989 Tiananmen Square military crackdown. Qwen3 returns an error message warning of 'inappropriate text' when asked about June 4, 1989, Doubao omits the protests and military intervention when listing historical events at the plaza, and DeepSeek responds 'I'm sorry, that does not fall within my current competencies' to indirect questions about the location.
The Chinese government has enforced decades-long censorship of any official review of the June 4, 1989 military crackdown that killed hundreds or thousands of student protesters according to independent organizations, though no official death toll has been released by Beijing. Chinese authorities intensify censorship especially around June 4 each year and in preceding days.
Reported impact
No mentions of Tiananmen events exist on Chinese social media platforms including Weibo and Douyin, the Chinese equivalents of X and TikTok.
Governance controls
The AI systems are deployed by major Chinese technology companies in a country where any public reference to the Tiananmen Square events has been strictly censored for decades following the incident. Chinese authorities cut off a major livestream influencer's broadcast in 2022 when he promoted a tank-shaped ice cream near the anniversary date, after which he disappeared from social media for three months.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.