“While the Chinese Communist Party won’t allow the Chinese people to use Twitter, it is happy to use it covertly to sow propaganda and disinformation internationally,” said Fergus Hanson, director of the International Cyber Policy Centre at the Australian Strategic Policy Institute, which issued a report Thursday analyzing the Chinese campaign. “Persistent, covert and deceptive influence operations like this one demonstrate the extent to which the party-state will target external threats to its political power.”
The campaign recently has broadened to exploit racial unrest in the United States, the ASPI found. One account, for instance, tweeted an image of Lady Liberty with a knee on the neck of George Floyd, the unarmed black man whose death last month after a white police officer in Minneapolis knelt on his neck has sparked global protests and calls for police reform.
Twitter’s disclosure that it had removed 23,750 accounts builds on an action this past August in which it removed other accounts that the social media firm explicitly linked with China’s ruling party.
The accounts were pushing the official, often-contested Chinese line on sensitive issues including the pro-democracy protests in Hong Kong (violent radicals encouraged by the United States), the coronavirus (the Chinese worked together and triumphed over it) and Taiwan (it learned its covid-19 response from China).
They composed the highly engaged core of a network that includes some 150,000 “amplifier accounts” that had few or no followers and were strategically designed to artificially inflate metrics to make it appear the tweets were highly popular, Twitter said.
These core accounts fired off 348,608 tweets between January 2018 and April 2020. Most were in traditional Chinese characters, as well as some special characters used only in Cantonese, which is spoken in Hong Kong. The propaganda campaign was aimed at Hong Kong residents, as well as the Chinese-speaking diaspora, researchers said.
“What we see in this data set is further evidence of a sustained commitment by the Chinese Communist Party to use social media for influence operations,” said Renée DiResta, technical research manager at the Stanford Internet Observatory, which also analyzed the data.
Twitter is blocked within China, where censors tightly control the Internet to try to ensure that the government version of events is the only one accessible. But over the past year, China has sought to broaden its online propaganda efforts beyond its domestic audience and has aggressively pushed its narrative in the wider world.
Chinese diplomats — called “wolf warriors” for their pugnacious style — have used Twitter to spread Beijing’s message to international audiences in English and other languages. The government also has begun to surreptitiously target Chinese-speaking audiences around the world using inauthentic or hacked accounts on digital forums such as Twitter and Facebook.
“This large-scale pivot to Western platforms is relatively new, and we should expect continued evolution and improvement,” given the government’s resources and interest in promoting its narrative, the ASPI wrote.
China’s Foreign Ministry did not answer The Washington Post’s request for comment on the core points of the report. Instead, it responded with a general call for countries to work together. “Both the United Nations and the World Health Organization have called on all countries to strengthen unity and cooperation in cracking down on disinformation,” the ministry said in a faxed statement.
In August, Twitter removed almost 1,000 accounts it said were linked to the Chinese government and focused on undermining the legitimacy of the Hong Kong pro-democracy movement.
Beijing responded by “immediately creating new [inauthentic] accounts to continue pushing the same themes, which include emergent priorities such as covid, and English-language content to surreptitiously push CCP talking points,” DiResta said.
Twitter’s takedown disclosures, she added, and their efforts to allow access to data for researchers, “offer a model for other companies to follow.”
The immaturity of the Chinese campaign is reflected in the low level of authentic user engagement — of real people reading and retweeting or commenting on the posts, and in the speed with which Twitter detected the accounts.
The operators favored speed and scale over quality, the ASPI noted. Content appeared to have been assembled hastily, with paragraphs squeezed in and images distorted. In some cases, the operators didn’t bother to remove spell-check underlines, the think tank said.
Most of the accounts tweeting about covid-19 were created after January, with one large batch emerging on a single day in April. Some tweets pushed the line that China was better at fighting the coronavirus than Taiwan, a self-ruled democracy that Beijing views as a breakaway province. “China is the best anti epidemic country in the world, not Taiwan,” tweeted a since-deleted account using the handle @NicoleS00264634.
The accounts also targeted billionaire Chinese fugitive Guo Wengui, who lives in Manhattan and is close to former Trump adviser Stephen K. Bannon. Guo, who also goes by Miles Guo and Miles Kwok, previously worked closely with Chinese intelligence officials but is now campaigning to topple the Communist Party. Beijing is seeking Guo’s extradition to face charges including fraud, blackmail and bribery,and once sent security agents to pressure him to cease his accusations of corruption against the party.
Some of the tweets depict Guo as a rat or show images of protesters outside his apartment holding bilingual signs saying “Guo Wengui is a big traitor.” Many focus on Guo’s relationship with Bannon, suggesting that Bannon is happy to be Guo’s mouthpiece for a price.
“The Chinese are so obsessed with their image, anybody criticizing them from abroad gets leveled with thermonuclear strikes,” said James Mulvenon, director of intelligence integration at U.S. defense contractor SOS International.
Twitter removed the accounts for violations of its policy forbidding platform manipulation, including engaging in deceptive activity or attempts to make accounts appear more popular than they are, and efforts to artificially influence conversations through fake accounts and automation.
“Improving the health of the public conversation is a priority for our company,” Twitter said in a statement. “If we can ever attribute these behaviors to a state-backed information operation, we disclose them to our public archive — the only one of its kind in the industry.”
Twitter traced the accounts to the Chinese Communist Party through specific unblocked IP addresses originating from the country, said a Twitter official, speaking on condition of anonymity to discuss a sensitive topic.
China is several years behind Russia, which has run online disinformation campaigns in Europe and former Eastern Bloc countries for years, and notoriously interfered in the 2016 U.S. presidential election with an effort to exploit societal rifts through posts on Twitter and Facebook that reached millions of Americans.
But while Russia seeks to divide societies and devalue the notion of truth, China, analysts say, is focused primarily on delegitimizing dissidents and adversaries and promoting an image of China as a world power.
“It’s still really easy to identify Chinese information operations because the Chinese are mostly interested in refuting criticism and presenting China in the best possible light,’’ said Mulvenon of SOS International. “The Russians play Bernie Bros against white nationalists, the left against the right and are much more technically sophisticated. The Chinese are not interested in devaluing truth because they think there is a truth to their message.”
Twitter and Facebook began aggressively to take down state-sponsored accounts pushing propaganda and disinformation after the 2016 election. Influence operations from Iran, China, Russia and several other countries have been detected by Facebook and Twitter in recent years.
Lyric Li in Beijing contributed to this report.