U of T's Citizen Lab exposes censorship on popular chat app, WeChat

Researchers at the University of Toronto’s Citizen Lab published a report today that reveals how WeChat, the most popular chat app in China, censors content.

The report shows WeChat has separate censorship policies for international users and those in China with the majority of censorship targeted for Chinese accounts. It also reveals that WeChat removed notifications to users about the blocking of chat messages on the platform. 

The findings are making headlines  here at home and around the world.

Read the Globe and Mail 

Read Bloomberg

Read The Wall Street Journal 

The researchers found that there is more censorship in “group chat” messages compared to one-to-one user chats, possibly due to concerns about posts being spread to larger audiences and leading to mobilization, and that WeChat’s built-in browser also blocks certain websites for both China and international accounts.

The researchers found 41 websites blocked exclusively for Chinese WeChat accounts, including online gambling, news and media websites that critically report on China and the website of the International Consortium of Investigative Journalists, which reported on the Panama Papers. 

“Attention usually focuses on foreign companies attempting to reach into China and facing hard decisions over how to approach its strict content regulations. WeChat has the opposite dilemma. To gain wider success the app must maintain its base in China, all while staying within the Chinese government’s boundaries, and present a compelling experience to attract international users,” says Masashi Crete-Nishihata, research manager at Citizen Lab, which is located at U of T's Munk School of Global Affairs.

Citizen Lab, which has extensive experience uncovering Internet censorship practices through network measurement and reverse engineering techniques, says the report shows the importance of understanding how apps work.

“Days are long gone when we used to interact with the Internet as an undifferentiated network,” says Professor Ron Deibert of the Faculty of Arts & Science, the director of Citizen Lab. “The reality today is that what we communicate online is mediated by companies that own and operate the Internet services we use.

“Social media in particular have become for an increasing number of people windows on reality. Whether, and in what ways, those windows might be distorted – by corporate policies or government directives – is thus a matter of significant public importance (but not always easy to discern with the naked eye).”

WeChat is the dominant chat application in China and fourth largest in the world with 806 million monthly active users. The application thrives on its huge user base in China, but like any other application in the country it must follow strict content regulations.

The report finds that WeChat enables keyword filtering for users with accounts registered to mainland China phone numbers. Remarkably, the researchers found that censorship stays on
even if users switch to a non-mainland phone number or travel to a different country – “locking in” users with mainland China accounts to its system of censorship no matter where they go.

“It’s unclear if the persistent content restrictions we've detected for China accounts is intentional, but the outcome is concerning. If you register a WeChat account to a Chinese phone number, you will always be under additional censorship even if you travel or later link your account to an international number. The idea that you can't escape a censorship system imposed on you at the time of registration is a troubling one indeed,” explains Jason Q. Ng, a researcher at Citizen Lab.

The researchers systematically tested a sample of keywords in two WeChat modes: one-to-one chat and group chat. They found a greater number of keywords blocked on group chat, which suggests that group chat is specifically targeted, potentially because of its ability to reach a larger numbers of users. Censored keywords spanned a range of content including current events, politics and social issues.

The report also found that censorship on WeChat is dynamic. Some keywords that triggered censorship in original tests were later found to be permissible in later tests. Newly censored keywords also appear to have been added in response to current news events.

“When you send a message on WeChat, it passes through a remote server that contains rules for implementing censorship. If the message contains a keyword or set of keywords that have been targeted for blocking, the message will not be sent,” explains Jeffrey Knockel, senior researcher at Citizen Lab.

The report goes on to detail how, in both one-on-one and group chat, censorship now happens without user notification. Previously if a user sent a message with a blacklisted keyword, a warning would pop up explaining the message could not be sent. Now messages are censored without giving any indication that they have been blocked.

“The removal of the censorship notices means WeChat has become even less transparent and also less dependable for its users in how it handles their communications,” says Citizen Lab researcher Lotus Ruan.

In addition to keyword censorship, WeChat implements a URL filtering system in its built-in browser. All of the sites that were exclusively blocked on Chinese accounts were fully accessible on international accounts without any warning page, but the researchers also found intermittent blocking of gambling and pornography websites on international accounts.

Unlike chat censorship, when a website is blocked on WeChat, a variety of explanatory messages are provided for why the censorship has occurred. However, it is unclear how accurately the purported explanations match up with the actual reasons for why websites are blocked. This ambiguity in attributing the source for the filtering again reflects the lack of transparency in how WeChat determines what “sensitive content” to block.