• SHARE

Flood global social media with fake accounts used to advance an authoritarian agenda. Make them look real and grow their numbers of followers. Seek out online critics of the state – and find out who they are and where they live.

China’s government has unleashed a global online campaign to burnish its image and undercut accusations of human rights abuses. Much of the effort takes place in the shadows, behind the guise of bot networks that generate automatic posts and hard-to-trace online personas.

Now, a new set of documents reviewed by The New York Times reveals in stark detail how Chinese officials tap private businesses to generate content on demand, draw followers, track critics and provide other services for information campaigns. That operation increasingly plays out on international platforms like Facebook and Twitter, which the Chinese government blocks at home.

The documents, which were part of a request for bids from contractors, offer a rare glimpse into how China’s vast bureaucracy works to spread propaganda and to sculpt opinion on global social media. They were taken offline after The Times contacted the Chinese government about them.

On May 21, a branch of the Shanghai police posted a notice online seeking bids from private contractors for what is known among Chinese officialdom as public opinion management. Officials have relied on tech contractors to help them keep up with domestic social media and actively shape public opinion via censorship and the dissemination of fake posts at home. Only recently have officials and the opinion management industry turned their attention beyond China.

Facebook headquarters in Menlo Park, California. Facebook said in September 2020 that it had detected limited Chinese operations intended to both help and hurt then-U.S. President Donald Trump’s re-election chances. | JIM WILSON/THE NEW YORK TIMES
Facebook headquarters in Menlo Park, California. Facebook said in September 2020 that it had detected limited Chinese operations intended to both help and hurt then-U.S. President Donald Trump’s re-election chances. | JIM WILSON/THE NEW YORK TIMES

Shanghai police were looking to create hundreds of fake accounts on Twitter, Facebook and other major social media platforms. The police department emphasized that the task is time sensitive, suggesting that it wants to be ready to unleash the accounts quickly to steer discussion.

Botlike networks of accounts such as those that the Shanghai police wants to buy have driven an online surge in pro-China traffic over the past two years. Sometimes the social media posts from those networks bolster official government accounts with likes or reposts. Other times they attack social media users who are critical of government policies.

Recently, Facebook took down 500 accounts after they were used to spread comments from a Swiss biologist by the name of Wilson Edwards, who had purportedly written that the United States was interfering with the World Health Organization’s efforts to track the origins of the coronavirus pandemic. The Swiss embassy in Beijing said Wilson Edwards did not exist, but the fake scientist’s accusations had already been quoted by Chinese state media.

The Shanghai police’s social media effort is not just a numbers game, and this portion of the document underscores efforts to shift from brute-force tactics like using bot armies to something more subversive:

Disguise and maintain overseas social media accounts. Suppliers should package a portion of the overseas accounts into a group of premium accounts, that is, accounts that survive for a long period of time, have a certain number of fans, and can be used to promote information. Each month on each platform three accounts must be maintained, and an increase in fans must be guaranteed each month. Note: this project has intermediate time sensitivity. Each week, the number of posts and survival rate of accounts will be calculated. If an account is suspended, it needs to be fixed in a timely fashion.

Protesters demonstrate against China’s growing political influence over Hong Kong, and interference with its autonomy, in the streets of the city on Jan. 1, 2020. Twitter said it has discovered and removed 23,750 accounts that were “highly engaged” in a coordinated effort to spread misinformation about protests in Hong Kong and the Communist Party’s response to the coronavirus, and that it also took down about 150,000 accounts that were dedicated to boosting China’s messages by retweeting and liking the content. | LAM YIK FEI/THE NEW YORK TIMES
Protesters demonstrate against China’s growing political influence over Hong Kong, and interference with its autonomy, in the streets of the city on Jan. 1, 2020. Twitter said it has discovered and removed 23,750 accounts that were “highly engaged” in a coordinated effort to spread misinformation about protests in Hong Kong and the Communist Party’s response to the coronavirus, and that it also took down about 150,000 accounts that were dedicated to boosting China’s messages by retweeting and liking the content. | LAM YIK FEI/THE NEW YORK TIMES

The police department was seeking an upgrade in sophistication and power: a series of accounts with organic followers that can be turned to government aims whenever necessary.

The request suggested that police officials understood the need for strong engagement with the public through these profiles-for-hire. The deeper engagement lends the fake personas credibility at a time when social media companies are increasingly taking down accounts that seem inauthentic or coordinated.

Bot networks that have been linked to China’s government stand out for their lack of engagement with other accounts, disinformation experts say. Though they can be used to troll others and boost the number of likes on official government posts, most of those automated accounts have little influence individually since they have few followers.

In their posting, authorities used a phrase common among China’s internet police that refers to tracking down the actual person behind a social media account: “touching the ground.”

Facebook, Instagram and WhatsApp apps on a smartphone in Hong Kong on July 7, 2020. | BLOOMBERG
Facebook, Instagram and WhatsApp apps on a smartphone in Hong Kong on July 7, 2020. | BLOOMBERG

With growing frequency, the country’s internet police has hunted down and threatened internet users who voice their opinions. At first, its agents focused on local social media platforms. In 2018, they began a new campaign to detain users of Twitter inside China – account owners who had found ways around the government’s blocks – and force them to delete their accounts.

Now, the campaign has extended to Chinese citizens who live outside of China. The document spelled out how Shanghai police wanted to discover the identities of people behind certain accounts and to trace their users’ connections to the mainland. Its officers can then threaten family members in China or detain the account holders when they return to the country in order to compel online critics to delete posts or even entire accounts.

In previous Chinese information campaigns, botlike accounts have been used to add an unrealistic number of likes and retweets to government and state media posts. The contrived flurry of traffic can make the posts more likely to be shown by recommendation algorithms on many social media sites and search engines.

In recent weeks, a similar pattern emerged from a network of botlike accounts amplifying evidence that was issued by state-media journalists, purporting to show that tennis player Peng Shuai was safe, freely eating dinner in Beijing and attending a youth tennis tournament.

Facial recognition technology is demonstrated on a video screen at the World Internet Conference is Wuzhen, China, on Nov. 7, 2018. Facebook plans to shut down its decade-old facial recognition system this month, deleting the face scan data of more than 1 billion users and effectively eliminating a feature that has fueled privacy concerns, government investigations, a class-action lawsuit and regulatory woes. | JONATHAN BROWNING/THE NEW YORK TIMES
Facial recognition technology is demonstrated on a video screen at the World Internet Conference is Wuzhen, China, on Nov. 7, 2018. Facebook plans to shut down its decade-old facial recognition system this month, deleting the face scan data of more than 1 billion users and effectively eliminating a feature that has fueled privacy concerns, government investigations, a class-action lawsuit and regulatory woes. | JONATHAN BROWNING/THE NEW YORK TIMES

Shanghai police explained very clearly the functionality that the department desires, demonstrating a familiarity with recommendation algorithms on social media. Its approach underscores something that propaganda officials know well: A cluster of junk accounts can briefly make one post from an official account appear to go viral, giving it greater exposure and lending it credibility.

As overseas Chinese propaganda campaigns have developed, they have come to rely more on visual media. Officials were looking for a company to not only maintain and deploy fake accounts, but to also generate original content. The demand for videos is high.

A separate document reviewed by The Times shows that the same local branch of Shanghai police purchased video-making services from a different company in November. Police asked the supplier to provide at least 20 videos a month and to distribute those on domestic and overseas social media. The document referred to the task as original video production that would be used to fight the “battle of public opinion.”

Earlier this year, a New York Times and ProPublica analysis showed how thousands of videos portraying members of the Uyghur ethnic minority living happy and free lives were a key part of an information campaign that Twitter ultimately attributed to the Chinese Communist Party. When Twitter took down the network behind those posts, it took down accounts linked to a contractor that it said helped shoot propaganda videos. A Twitter spokesperson declined to comment.

Three weeks after the Shanghai police department’s request became public, a company called Shanghai Cloud Link won the bid, the documents show. In its pitch, the company listed itself as having just 20 employees. According to the LinkedIn page of its founder, Wei Guolin, the company works with multinational firms and provides services in “digital government” and “smart cities.”

Wei did not respond to a request for comment. The Shanghai Pudong Public Security Bureau did not respond to a faxed request for comment.

Work like what Shanghai Cloud Link pitched is likely just the tip of the iceberg. Local governments and police across China have put out similar requests for services to influence overseas social media but often in vague terms. Occasionally, specifics are revealed. In 2017, for instance, police in Inner Mongolia purchased software that allowed government trolls to post directly to multiple social media sites, inside and outside of China, according to documents reviewed by The Times.

In another case, a contractor had downloaded hundreds of access credentials for Facebook’s public feed, allowing it to collect data about who commented on which posts and when. Facebook did not immediately comment.

Shanghai Cloud Link’s winning bid offers a window into how much some of these types of disinformation services can cost.

In many cases, tech contractors seek to sell Chinese authorities the hardware and software outright. In this case, Shanghai Cloud Link’s proposal hinted at a new service-based model, one in which officials pay on a month-by-month basis – a sort of subscription for social media manipulation.

© 2021 The New York Times Company
Read more at nytimes.com

In a time of both misinformation and too much information, quality journalism is more crucial than ever.
By subscribing, you can help us get the story right.

SUBSCRIBE NOW

PHOTO GALLERY (CLICK TO ENLARGE)