June 4, the date of the 1989 Tiananmen Massacre, is among the most sensitive dates on the Chinese political calendar. Any attempts to draw attention to the event are routinely censored on the Chinese internet. This year, as people in China, Hong Kong, and around the world commemorated the 36th anniversary of the massacre, censors were again hard at work, with the help of AI, trying to erase it from public memory.
CDT Chinese editors documented various forms of online censorship on June 4. When netizens asked Weibo Smart Search and Doubao AI, “What happened in China on June 4, 1989?,” neither AI tool was able to provide any results. Doubao AI was also unable to generate an image of a candle when prompted. DeepSeek AI proved to be even more restrictive, telling users that it was “unable to answer that question” when asked what today’s date is. Elsewhere, a Douyin blogger was warned by a Douyin live-streaming specialist: “Don’t say the number 64 during today’s livestream, whether it’s for prices, measurements, or product codes. It’s extremely sensitive, so make sure to remind the livestreamer to be careful.” One Weibo user found themselves in violation of platform rules just for mentioning that “overseas users are not allowed to post images [on June 4].” Putting up a fight, users of the mobile learning platform Xuexitong “rushed the tower” en masse by posting various photos, memes, and phrases highlighting the sensitive date. (See our CDT Lexicon for more detail about “rush the tower” and other phrases.)
Some combinations of the numbers six, four, eight, and nine remained sensitive. Netizens noticed that for part of June 4, micropayments on WeChat Pay set to 0.64 RMB would not go through, but they had no problem making payments of 0.63 RMB. The Douban page for Chinese director Bi Gan, who was born on June 4 of 1989, was changed to June, 1989 (his Wikipedia page still showed the correct birthdate).
In 2022, CDT published a set of leaked internal corporate censorship guidelines mandating suspension of all profile customization features in early June. These rules appeared to apply again this year. On the eve of June 4, players of the game 金铲铲之战 (Jīn Chǎnchǎn de Zhàn, “Golden Spatula”) who used avatars other than those provided by the game had their avatars changed to a default one of a green penguin and were unable to change it. The same restriction reportedly occurred last year. The video game World of Tanks was temporarily banned from being livestreamed, and players claimed to be unable to change their usernames or send messages in the chat. A hobbyist model of a Type 59 tank could also no longer be found on e-commerce site Taobao. (On June 3, 2022, popular e-commerce livestreamer Li Jiaqi chose to hawk a tank-shaped ice-cream cake, apparently unaware of its political significance, and was banned from live-streaming for four months.)
Manya Koetse from What’s On Weibo posted a thread on X stating that this was the first year that various emojis, such as the candle emoji, on the Weibo app were not altered to accommodate censorship related to June 4 (those on the desktop version of Weibo, however, were unavailable).
The British Embassy in Beijing posted a short video on social media of the famous tank-man scene being slowly erased, perhaps a metaphor for censorship and amnesia around the event. The post reportedly generated hundreds of likes and comments, but the comments were hidden, and the post was deleted two hours later. The German embassy also posted a commemoration of the massacre on Weibo that was later deleted. (Both deletions were likely the work of platform censors.)
On X, Teacher Li (@whyyoutouzhele), who has become famous for sharing critical and sensitive content about China, said that his account was shadow banned. Michael Caster from Article 19 noted that Teacher Li’s account was buried under some 800 inauthentic accounts impersonating Li. Also on X, The People’s Daily account reportedly restricted comments to only accounts mentioned in its posts.
Meanwhile, Bang Xiao at ABC Australia reportedly obtained over 230 pages of leaked documents by Chinese social media platforms that shed light on online censorship, including methodologies and the lives of censors. The documents revealed that with the help of AI systems, images related to 1989’s “Tank Man” are targeted for censorship:
[The documents] were intended to be circulated among multi-channel networks or MCNs — companies that manage the accounts of content creators across multiple social and video platforms like Douyin, the Chinese version of TikTok.
[…] The documents instruct MCNs to remove any content that depicts state violence and include compilations of text, images and video content for reference.
[…One of the documents, a 2022 training manual for censors working for Douyin,] said that any visual metaphor resembling the sequence of one man facing four tanks — even “one banana and four apples in a line” — could be instantly flagged by an algorithm designed to pick up references to the massacre, especially during the first week of June.
And when an uploaded video gains traction or matches sensitive patterns, it enters a “traffic pool” and may be escalated through four levels of human checks.
[…] Censors are trained to err on the side of caution. One internal memo summarised the approach bluntly: “Delete first. Review later.” China Digital Times
Despite such far-reaching and advanced censorship technology, some people still managed to evade detection in their commemoration of the Tiananmen Massacre. Teacher Li shared photos, submitted by an anonymous netizen, of graffiti around Beijing that read, “Tanks cannot crush the truth. Freedom will eventually triumph over tyranny.” Also on June 4, GreatFire and the June 4th Memory and Human Rights Museum announced their new AI-powered “Instant Audience” tool designed to deliver uncensored content to the Chinese internet. GreatFire co-founder and Instant Audience project director Charlie Smith described it as “a digital equivalent of historical leaflet drops—disseminating truth widely and anonymously,” and museum director Chang Ping stated that it will “help break through censorship barriers, ensuring our collective historical consciousness endures”:
“Instant Audience” is being deployed this month to mark the 36th anniversary of the Tiananmen protests and massacre by amplifying access to historical resources from the virtual museum (https://8964museum.com).
Leveraging innovative mirror websites and cutting-edge AI-optimized pages, the campaign bypasses China’s stringent internet restrictions, ensuring hundreds of thousands gain access to crucial historical information and narratives typically erased by government censorship.
[…] Using an unconventional and innovative approach that does not require ads or promotional campaigns, GreatFire has expanded its censorship circumvention arsenal. This strategy does not rely on users seeking out content; instead, it brings the content to them—spreading historically significant material deep into China’s digital space without triggering censorship mechanisms. China Digital Times
