Everyone knows that China has some of the most sophisticated censorship tools in the world, but the details of how they actually work — what they censor and when — are often not fully understood. A new report by Citizen Lab, a research group studying the web, human rights, and global security, sheds some light on one particularly fruitful target for Chinese censorship: mobile messaging.
Citizen Lab looked at how the Chinese government censors discussion on WeChat, a popular messaging app. WeChat is the fourth biggest messaging service in the world, with more than 768 million active users, but is also deeply embedded in Chinese society, where it’s used not only for chatting, but for tasks like banking, paying bills, booking holidays, calling cabs, and much more.
The cornerstone of WeChat censorship is keyword filtering, which blocks messages that contain terms like “human rights,” “mass arrest,” and “spiritual freedom.” However, Citizen Lab found that the censors don’t just block messages containing any one specific phrase, but instead look for combinations of different terms. So you can send a message with the words “human rights lawyer” in it, but if you combine that with the name of a specific lawyer — Jiang Tianyong, who was recently “disappeared” by the government — the message is blocked.
When a message is censored, users are not notified of this fact. They see it as sent in their own app, but it just never reaches its intended recipient. The system works by examining every message that is sent when it passes through WeChat’s servers. The list of filtered keywords is also reactive, and changes in relation to the news; and only to WeChat accounts using mobile phone numbers registered in the Chinese mainland. Citizen Lab says much of the censorship on WeChat is currently focused around the “709 Crackdown” — a series of arrests against civil dissenters that began on the 9th of July 2015 (hence the name).
An interesting quirk of WeChat censorship discovered by Citizen Lab is that it’s stricter when it comes to group discussions. The group found that more keyword combinations were blocked in chats containing multiple users than in one-on-one conversations. The reason for this isn’t clear, but it could be the Chinese government thinks it prudent to allow limited discussion of sensitive topics, but that group conversations are more dangerous, perhaps leading to organized dissent. WeChat Moments (a feature similar to Facebook’s News Feed) was also more heavily censored, with certain images filtered out as well.
The report notes: “The greater attention to group chat and Moments in particular may be due to the semi-public nature of the two features. Messages can reach and inspire discussions among wider audiences, making it subject to a higher level of scrutiny.”
For a full list of censored keywords and combinations, you can read Citizen Lab’s report in full here.