Chinese mass influence attempt ‘way bigger than we thought’

A freshly uncovered mass manipulation effort from China was far greater and more widespread than anyone anticipated.

The accounts portrayed Hong Kong protesters as violent and as cockroaches. Picture: AFP

A freshly uncovered mass manipulation effort from China was far greater and more widespread than anyone anticipated, according to the Australian who helped unmask the government-sponsored campaign.

Twitter said on Friday it had removed more than 170,000 ‘bot’ state-linked accounts tied to China, as the tech giants continue to grapple with widespread fake news, misinformation and manipulation flooding their platforms.

A team of Australian researchers from the Australian Strategic Policy Institute (ASPI) found that China’s influence operation targeted Chinese users outside of the mainland – where Twitter and Facebook are blocked – aiming to manipulate opinion on issues including the Hong Kong protests, the coronavirus and Taiwan.

The ‘bot’ accounts tweeted at working hours at Beijing time, with posts spiking at 8am to 5pm Beijing time and dropping off on weekends, and most accounts had fewer than 10 followers and no bios.

The accounts portrayed Hong Kong protesters as violent and as cockroaches, accused exiled Chinese billionaire Guo Wengui as being immoral and a rapist, and aimed to deflect away from China’s role in allowing the coronavirus to spread.

Australian Strategic Policy Institute senior analyst Dr Jacob Wallis said this represented a ‘new normal’ when it comes to government espionage and propaganda.

“This is an attempt to manipulate social media audiences at scale; it’s way bigger than we thought,” Wallis said.

“The challenge is that this is an adversarial battle with malign actors who will play a game of cat and mouse with the platforms, and are constantly refining their techniques and tactics. It’s going to be a constant to and fro and that’s the challenge if we want to have an open internet.”

Wallis added that China had largely failed in that the bot accounts were effectively a self-contained echo chamber, and hadn’t found a way to break into more authentic Twitter networks.

“What they’re doing is managing to create new accounts in order to keep the operation rolling on, 95 per cent of these accounts had less than eight followers, so they’re really fresh accounts.”

He predicted the manipulation efforts would intensify as the US election draws closer in November.

Australia or the US would not be able to retaliate with a similar effort Wallis said, given China removes political viewpoints, topic or individuals it disagrees with. Weibo – China’s version of Twitter – routinely blocks legitimate diplomatic posts from foreign embassies, for example.

“If we look at other countries who are more willing to simply block anything they don’t wish their citizens to engage with in any way … That’s not the kind of internet we would like to see. We value openness, the sharing of opinions, and the capacity to critique and engage with divergent viewpoints.”

Twitter has publicly released an archive of 23,750 suspended accounts that are connected to state-backed information operations, while a further 150,000 haven’t been released due to privacy reasons.

Wallis said that was likely because those 150,000 accounts had been hacked, stolen or bought on the dark web, and likely previously belonged to a legitimate user.

The research will intensify calls for Twitter to kick off all Chinese Communist Party (CCP) officials from its site. In March, US Republicans Ben Sasse and Mike Gallagher complained to Twitter CEO Jack Dorsey in a public letter urging him to ban CCP officials.

The CCP was “waging a massive propaganda campaign to rewrite the history of COVID-19 and whitewash the Party’s lies to the Chinese people and the world,” they wrote.

Paul Barrett, deputy director of the Center for Business and Human Rights at New York University, recently authored a report into the tech giants’ moderation of content on their platforms. He told The Australian this discovery was likely the tip of the iceberg when it comes to hidden bot propaganda networks on platforms like Twitter.

“Twitter and the other platforms ordinarily detect this kind of large-scale influence campaign by means of automated systems driven by artificial intelligence,” Barrett said.

“An optimist would see a social media company aggressively patrolling its platform and weeding out state-sponsored disinformation. A pessimist would see the Chinese growing more assertive in imitating covert Russian techniques and speculate that if Twitter found this network, there are probably many others out there that haven’t been detected.”

A Twitter spokeswoman told The Australian that Twitter was the only social media company to publicly release its data.

“Our archive is the only one in the industry and we’ve been disclosing all state-linked activity we can reliably attribute since October 2018,” the spokeswoman said. “We focused on behaviours, not content. We’re not an intel agency and don’t infer all political intent — we remove violative accounts, attribute where possible, and disclose publicly so independent analysis from the experts can be conducted.”

The spokeswoman added that more than 1 in 2 of tweets we take action on for abuse are proactively surfaced using technology, rather than relying on reports to Twitter, compared to 1 in 5 tweets in 2018.

“If people on Twitter see something that may violate the Twitter Rules, the most important thing they can do is to report it. Reporting helps signal to us that something is going on, and it helps us keep conversations on Twitter healthier in the long term. We can take action on individual tweets, Direct Messages, or accounts.”

A Facebook spokeswoman said the network from China removed by Twitter was connected to a spam network Facebook previously removed in 2019 and 2020.

“We have been closely monitoring for any efforts to re-establish its presence on Facebook,” she said. “As part of this work, we’ve also removed some additional spam accounts this week.”