WeChat “monitors text messages from international users to preventively identify content to be censored for its Chinese users”
- The Citizens Lab studied how WeChat builds its database of blacklisted terms
- They discovered that the company was monitoring international messages to preventively identify potential images or content that should be banned in China
- The application creates a “hash” code for potentially offensive messages and adds the hash to a database that will block the sending of these images to China.
- The system allows the application to censor certain subjects without even being asked
WeChat monitors messages sent by international users to help identify content that may need to be preemptively censored in China.
The popular text messaging application maintains a database of censored or potentially inflammatory content which it uses to automatically filter messages from Chinese account holders.
The company keeps this database up to date by scanning messages sent between account holders outside of China to preemptively mark new images or other content that could be considered inflammatory by the Chinese government.
Popular chat app WeChat would monitor international chat exchanges to identify and label content to be censored for Chinese account holders
The system has been documented in recent tests by The Citizens Lab, a Canadian institute of public policy at the University of Toronto.
“I urge international users to consider that when you use this platform, you are actually helping to strengthen digital repression in China,” said Citizens Lab director Ron Deibert. South China Morning Post.
Previous reports have shown that WeChat maintains an active blacklist of around 500 keyword combinations that will be automatically blocked in messages with Chinese account holders.
For example, a message mentioning Dr. Li Wenliang, an ophthalmologist from Wuhan Central Hospital credited as a COVID-19 whistleblower, will likely be blocked if the intended recipient is a Chinese account holder.
To try to learn how WeChat is building this blacklist, they tested how the app responded to their attempts to send a new illustration of Liu Xiaobo, a jailed writer and activist sometimes referred to as “ Nelson Mandela of China ”
According to The Citizens Lab, the app maintains a database of blacklisted keyword combinations and an index of hash codes attached to images or files that should be censored. Hash codes seem to come, in part, from the analysis of international messages
In a message thread with only account holders outside of China, they sent a picture of Liu Xiabo who arrived intact, but WeChat assigned him a special hash code.
When the researchers sent the same message to a thread including a Chinese account holder, they discovered that WeChat was using the hash code generated from their previous thread of international users to identify and delete the image. .
They used image editing software to modify the content of the image itself in several ways, but made sure it always had the same hash code in the image file, but WeChat always blocked automatically image for Chinese users based only on the hash code.
The Citizens Lab tested the functionality by sending a new illustration of the imprisoned Chinese dissident Liu Xiaobo to a group of international users. They found that WeChat generated a new hash code for the image and used it to censor the image when it was later sent to a Chinese user.
“If the users [of WeChat] weren’t previously concerned, they should be very concerned now and reassess the risks of using this app, “said Deibert.
In a statement to the South China Morning Post, a WeChat spokesperson said the company takes the report seriously, but that regarding the suggestion that we engage in monitoring the content of international users, we can confirm that all content shared between international WeChat users is private. ”
The company did not respond to any of the report’s specific claims, but said, “Our policies and procedures are in accordance with all laws and regulations of each country in which we operate”.
A WeChat spokesperson said they take the Citizen Labs report seriously but stressed that “our policies and procedures comply with all laws and regulations of each country in which we operate”
According to Sarah Cook, research analyst at the Freedom House think tank, the report has disturbing implications.
Cook warned that WeChat tools could be used to “identify certain users and create a portfolio about them, [or] nurture other aspects of the [Chinese Communist Party’s] transnational repressive apparatus. “
“This will amplify calls for further examination, particularly in the United States, of WeChat,” said Cook.