Published by The Lawfare Institute
in Cooperation With
In early October, the Twitter account of a vice president at an international private bank based in Monaco tweeted several times about life in Xinjiang, China. She shared videos of the white birch forests that beautify the region and spoke of her love of Xinjiang-style fried pork noodles. The account also posted a video of the Xinjiang cotton harvest, saying, “Mechanization helps the cotton industry improve quality and efficiency, and increase farmers’ income #Xinjiang.” This banker had not recently returned from a trip to Xinjiang. She was not tweeting praise for Xinjiang culture and economy out of genuine affection or self-interest. She was not even doing the tweeting. Hackers had stolen her account some weeks before to join a chorus of other Twitter accounts discussing Xinjiang’s cuisine and cotton. This banker’s Twitter account had become the smallest cog in a vast, state-backed, defensive-disinformation campaign.
In the past several years, inauthentic social media accounts attributed to China have been identified as operating as part of several disinformation campaigns. China’s trolls first made headlines when they worked to undermine Hong Kong democracy protests. Since then, actors linked to the Chinese government have expanded the use of similar accounts. The cybersecurity firm FireEye and Google’s threat analysis group have identified an ongoing Chinese information operation involving social media. They have identified elements of this campaign in various languages across 30 different social media platforms and 40 different websites and found it targeted a range of issues, including attempts to undermine the dissident Chinese billionaire Guo Wengui. More recently, researchers from Oxford University’s Programme on Democracy and Technology revealed China-linked accounts that spread an absurd story that the coronavirus originated with Maine lobster shipped to Wuhan.
Our research at the Clemson University Media Forensics Hub has been tracking an extensive pro-China propaganda and disinformation operation, extending the work done by FireEye, Google and Oxford. Understanding the full strategy, tactics and motivation behind current and future Chinese disinformation campaigns requires an understanding of the full Chinese tool set.
China, generally with a broader range of critics than allies, faces serious political and public relations challenges, from conflict over the South China Sea to Hong Kong to Taiwan. With the approach of the 2022 Beijing Olympics, China has developed an extensive, multifaceted media apparatus to compete for influence on the world stage to combat these critics. The country’s capacity to smooth over its PR problems by defensively projecting its own version of reality into the media ecosystem may be more important than ever. One issue challenging China’s global image is the Chinese government’s treatment of the Uyghur Muslim minority in the northwestern region of Xinjiang. Human rights groups have accused China of detaining more than 1 million Uyghurs in state-run “reeducation camps” where they are victims of forced labor—and perhaps even forced sterilization. Amnesty International has labeled China’s actions as “crimes against humanity.” China has responded to these accusations with propaganda and disinformation, as these criticisms threaten the nation’s economy.
Cotton is an important Chinese export. As the world’s second largest producer of cotton, China is close behind its rival India. Xinjiang accounts for 85 percent of Chinese cotton production and 20 percent of total world supply. Xinjiang cotton is used by many of the world’s best known fashion retailers, including H&M, Ralph Lauren and Calvin Klein. Human rights organizations say the fields and factories of Xinjiang are run using forced labor from Uyghur Muslims, but industry efforts to motivate reform so far have been silenced by concerted Chinese economic pressure. Global attention remains focused, however, and the voices of those critical of China’s treatment of the Uyghurs won’t be swayed by brute economic measures. Threats to boycott Xinjiang cotton persist, so China has ramped up its communication and disinformation machine to drown out criticism.
If the Twitter account from Monaco is a small cog in China’s communication machine, the central mechanism is state media. China engages in a great deal of traditional propaganda driven by a range of state-controlled media outlets that push content in several languages across every form of media. These outlets have worked to undermine accusations of human rights violations made by “anti-China forces.” China Daily, in one of many such examples, suggested that accusations of abuse are based on fake testimonies and fabricated documents, quoting Chinese spokesperson Xu Guixiang: “As more and more people learn about the real situation in Xinjiang, they can see clearly that those allegations are made up to curb China’s development.”
Caption: A tweet from an inauthentic Twitter account undermining the accusations of an "anti-China force."
Chinese media have been working diligently to construct and portray “the ‘real’ situation in Xinjiang,” and many of these efforts have been well documented by Western media. With 12.3 million followers, the Chinese Xinhua News Twitter account exemplifies this, posting in English twice a day about Xinjiang. Xinhua News shares Travel Channel-worthy drone shots of breathtaking landscapes alongside video that invites the viewer to experience the food and culture of the region. One recent video invites users to go shopping “with a young Xinjiang lady and her friend in this #RealLifeXinjiang #vlog.” Another video introduces users to Zufati, a 6-year-old who is the “fruit of love of a mixed-race marriage” and whose mother is lovingly teaching him Mandarin. And, of course, Xinhua News wants users to understand how thoroughly mechanized the Xinjiang cotton fields are. Xinhua tells in one post, for instance, how Xinjiang farmers are celebrating the cotton harvest, saying, “Watch how smart technologies help them free from arduous physical labor and increase income.”
State media’s messaging is supported by other official government Twitter accounts. Zhao Lijian, deputy director of the Ministry of Foreign Affairs, messaged his 1 million followers an average of once a day in English about Xinjiang throughout October and early November 2021. His feed is filled with videos of Xinjiang’s primary school children at play and of Xinjiang’s elderly citizens dancing in the streets. “Is this what you called ’genocide‘?” proclaims one such tweet. And, of course, many of his tweets are about Xinjiang cotton.
Alongside these overt propaganda efforts, China has invited foreign journalists and other guests to view the Xinjiang province firsthand. Some Western journalists have come away from these trips clearly very skeptical. Others, however, have taken up the party’s line and are eagerly magnified by state media. Xinhua News told of one Italian’s journey, quoting him saying, “I do see a situation which is pretty much normal and similar to the rest of China.” A Ugandan Muslim journalist had his story of religious acceptance in Xinjiang told by Chinese state channels and across social media.
In addition to individual journalists, China has recruited entire publications to help propagate its messages. The Helsinki Times, for instance, which proclaims in its own header to be “News from Finland,” has a large portion of its content supplied in English “in cooperation” with China’s People’s Daily—likely often unbeknownst to the reader.
Caption: The disclaimer on the Helsinki Times "China News" header.
With a swarm of inauthentic social media accounts, China’s efforts move from old-fashioned propaganda to 21st century disinformation. The ways in which China employs inauthentic troll accounts in this case, however, differ from what is typically seen in the context of Russian troll accounts. Analysis of Russian disinformation has shown that, in the past, Russian trolls have used social media offensively, attacking the West by integrating themselves into identity groups and working to undermine institutions while pulling conversations in more extreme directions. The Chinese trolls talking about Xinjiang, and arguably most troll accounts from past Chinese campaigns, are defensive trolls. They don’t work to attack the West; rather, they attempt to defend China’s interests.
Our research has been monitoring conversations around the hashtags employed by users engaged in discussions of Chinese treatment of the Uyghurs. Sampling these messages, we can see clear signs of platform manipulation. On Twitter, troll accounts engaging in these conversations are specialized and serve a variety of functions. First, and most obviously, they spam conversations by repeating talking points and filling space on the platform. It is difficult to pin down exactly how many accounts across social media the Chinese employ, but they number in the thousands. We found that the hashtag “#Xinjiang” appeared in 60,516 tweets in October. Of these, 3,310 tweets were the first tweet a given account ever made, suggesting the possibility these 3000+ accounts were purpose-built. Nearly 40 percent of these 60,000+ tweets originated from accounts that had zero followers, another sign of potential inauthenticity. The disinformation researcher @conspirator0 identified that many of these accounts were created simultaneously, in batches.
These accounts mostly copy and paste content, much of which discusses Xinjiang and, naturally, Xinjiang cotton. A typical post is like this recently suspended one from an account named Tanya Williams. It includes a 13-second video of a cotton field in Xinjiang and shares the message “#humanrights #cotton #xinjiang #forcedlabor #uyghur A good harvest of beautiful cotton.” Users searching for conversations using any of those hashtags are somewhat less likely to find content critical of China and a little more likely to land on a video of a cotton field. These tactics make it incrementally more difficult for users’ messages expressing genuine concerns about, or evidence of, China’s actions in Xinjiang to break through.
Caption: A pro-China tweet from the, now suspended, Twitter account of Tanya Williams.
This tactic of flooding a hashtag is akin to what is called “barrage jamming” in the context of electronic warfare—blinding a system by filling the display with noise. On social media, flooding a hashtag is a tactic made famous by Korean boy-band fans in June 2020 when they took over the use of “#whitelivesmatter.” K-pop fans did this as an anti-racist act, hoping to ensure white supremacists could not organize around the hashtag. China’s hashtag flooding is not similarly altruistic.
Another important function these accounts serve is that they frequently retweet posts from Chinese state officials, giving these government channels the appearance of legitimacy offered by high retweet counts. It’s important to remember, China doesn’t allow its citizens to freely use Western social media platforms.
Most of the Chinese troll accounts are not intended to stand up to scrutiny. They have handles such as “@HeidiBI79426365” and “@Brittan67692139.” Their profiles are poorly crafted, and their English is weak and repetitive. They are not created to last. As of Nov. 1, Twitter had suspended about 75 percent of the newly created accounts that operated in October. But these accounts’ purpose has been fulfilled, and the accounts themselves are expendable. The hashtag was flooded; the retweets and likes were given (and remain).
Caption: A tweet from the pro-China account called "Hitler 2" and the repost from an account called "garyhorton," both with either 0 or 1 followers/following.
But the banker from Monaco was not one of these newly created throwaway accounts. Her account had existed and been active for several years, although her interest in Xinjiang was relatively newfound, starting only weeks before Twitter suspended the account at the end of October. And she wasn’t alone. There are hundreds more accounts with profiles that looked much like hers—a book translator from France, a defunct business in New York City and a Korean comic book fan. These accounts were older, spread all over the world and were mostly dormant before acquiring a sudden interest in Xinjiang economic development (preceded by, in some cases, strong negative opinions about Steve Bannon, Guo Wengui and Li-Meng Yan). A little digging revealed that many of these accounts shared one further feature—a username that appeared in an email address included in a database of hacked accounts maintained by DeHashed.
These accounts played a third role in this propaganda operation, halfway between the raw burner accounts and the official spokespeople like Zhao Lijian. More difficult to identify relative to their newly created co-conspirators, they give a veneer of authenticity to their messaging. These less immediately suspicious accounts were often supported in their messaging by a phalanx of retweeting and liking from brand-new burner accounts to make them look even more legitimate, forming a small inauthentic propaganda cell.
China’s propaganda efforts around its treatment of the Uyghurs and the country’s attempts to undermine Western criticisms of its atrocities are well known and still ongoing. China is driven to maintain face and, as its focus on cotton suggests, defend its economic interests. The social media disinformation campaign we have described here, however, illustrates the steps China is willing to take not only to propagate its own version of reality but also to undermine others’ attempts to engage with the truth. China’s troll army not only reinforces China’s narrative but also limits the ability of journalists, activists, and victims to organize and employ social media as the tool of free speech it was once lauded for being. As China’s propaganda and disinformation networks adapt, those who try to break through the chaff will need to as well.