Facebook has removed two networks for ‘coordinated inauthentic behaviour’, one of them originating from China
An investigation by Facebook and a partner analytics firm has detailed how a network of social accounts from China tried to interfere with online political debate on hot-button topics ranging from the South China Sea to US politics.
Internet experts said the findings indicate an increasingly forthright attempt by Beijing to shape global discussion on the country’s policies and its relations with others, regardless of whether these tactics actually work.
“China’s propaganda campaigns conducted via overseas social media are becoming more aggressive and hawkish than ever,” said Fang Kecheng, an assistant professor of journalism at the Chinese University of Hong Kong.
Facebook said on Tuesday it had removed two networks for “coordinated inauthentic behaviour”, one of them originating from China. A total of 115 Facebook accounts and six Instagram accounts were taken down.
While these accounts allegedly used virtual private networks (VPN) to hide the source of their traffic, Facebook said it managed to trace them to the southern Chinese province of Fujian. The company did not elaborate on how they tracked down the location of the accounts and whether they were tied to the Chinese government.
“The networks taken down don’t seem very effective because based on the announcement, the majority of [these pages] basically just replicate among the same group of people, and some of the pages don’t have any followers,” said Fu King-wa, associate professor at the Journalism and Media Studies Centre at the University of Hong Kong.
Fu said that China has become “more proactive in disseminating its narrative in the international domain” on political issues but noted that this report did not explicitly mention the Chinese government as the source of the networks.
This is the first time that Facebook has taken down Chinese-based accounts for foreign interference and engagement in US politics, said the company’s cybersecurity policy chief Nathaniel Gleicher, according to a Reuters report.
The Chinese-linked groups and pages were often affiliated and shared administrators, a sign that they belonged to the same network, said Graphika, a social analytics firm that collaborated with Facebook on the report, which is called
Their efforts began in 2016 with a Chinese-language page called “Things about the Taiwan Strait”. Researchers found that at first, the Chinese-based accounts posted innocuous content like panda videos. But they soon started creating another page and posting comments that touted China’s power and attacked Taiwan President Tsai Ing-wen.
By 2018, the Chinese network had built two new pages devoted to discussing naval activities in the South China Sea. The posts defended Beijing’s policies in the highly-contested waters that are claimed by multiple countries. The US was often criticised. But at some point, they deviated to other themes.
Around eight months before the 2018 Indonesian presidential election, one group began posting content supportive of President Joko Widodo and blasting his main rival Prabowo Subianto. The posts stopped four months before Indonesians went to vote, when the group returned to its original focus on the South China Sea.
The Chinese-linked accounts appeared to adjust their strategy according to current affairs. When anti-government protests spread in Hong Kong last year, the groups pivoted to censuring demonstrators.
But none of these groups and pages were as successful in capturing attention as the ones dedicated to the Philippines, according to the report. They praised President Rodrigo Duterte, who until recently has
, as well as his allies. Two of those pages drew around 40,000 and 57,000 followers, researchers said.
In comparison, one South China Sea page attracted 16,000 followers. No other pages had more than 10,000 followers. When the Chinese-based operation expanded to cover American politics, one group focused on President Donald Trump and another on presidential hopeful Pete Buttigieg, drew only a handful of members.
One that posted screenshots of anti-Trump tweets had no followers at all.
Sometimes, the open nature of Facebook led to unintended consequences. Researchers said one South China Sea page drew real American users who posted anti-China messages.
“[China’s efforts] seem to be not very successful … as they face pushback from Twitter and Facebook as well at this stage,” said Yim Chan Chin, a media and communications professor from Xian Jiaotong-Liverpool University in Suzhou.
She said the Chinese government has outsourced projects to companies overseas to help build up an audience on western social media. But these companies often use questionable tactics like buying fake accounts, which violates Facebook and Twitter policies.
Others said that many of these efforts weren’t really designed to be compelling propaganda. Rather, they were carried out to please Beijing.
“For them, it is unimportant whether or not these campaigns are effective,” said Fang. “The purpose of these campaigns is to show loyalty to Beijing. They care more about what Beijing thinks rather than what foreigners think.”
Researchers from Graphika suggested that the Chinese operation might want to infiltrate online communities on both sides of the political spectrum. Facebook found posts in Southeast Asia that both praised and criticised China, while content in the US also showed divergent views on Trump and Joe Biden.
Despite the mixed success of their efforts, the Chinese-linked accounts took care to mask their identities and locations with increasingly sophisticated technologies, according to the report.
Initially, they took Instagram pictures and other widely available online photos as their profile pictures. Even though they may look legitimate at first, a simple process of reverse image search would reveal where the images originally came from.
The URL of their profiles also revealed that they changed their names after signing up for their Facebook accounts. For example, researchers found that the profile of a Zhou Xiaojing showed the name Johnson Martinez in the URL.
Later in their operations, some Chinese-linked accounts turned to artificial intelligence. They used faces created by a type of AI called generative adversarial networks (GAN).
These images of fake people are not perfect – the human eye can usually tell that something is off. To solve the problem, the Chinese accounts cropped some photos or posted stickers on them to hide any flaws, researchers said.
Tactics like these are not exclusive to campaigns in China, though, said analysts who called for Facebook to be more open about its content policies.
“The Facebook campaign does not target China in particular, since previously it has also removed accounts from Russia and the US as well,” said Fang. “However, there’s a lack of transparency in the mechanism behind the bans, and more clarity is needed on the way the social media platform handles data and regulates content.”
“[Facebook needs] to have a consistent policy and standards towards disinformation which apply across the different parties, not only because you’re coming from China,” said Chin.