A Chinese bot network played a key role in spreading disinformation during and after the US election, including a debunked video of “ballot burning” shared by Eric Trump, a new study reveals.
The misleading video shows a man filming himself on Virginia Beach, allegedly burning votes cast for Donald Trump. The ballots were actually samples. The clip went viral after Trump’s son Eric posted it a day later on his official Twitter page, where it got more than 1.2m views.
The video was believed to have originated from an account associated with the QAnon conspiracy theory. But the study by Cardiff University found two China-linked accounts had shared the video before this. Twitter has since suspended one of them.
The same Chinese network has spread anti-US propaganda, including calls for violence in the run-up to the 6 January storming of the US Capitol building by a pro-Trump mob. Afterwards. It compared the west’s response to the DC riot to political protests in Hong Kong.
The accounts previously posted hostile messages about Trump and Joe Biden, made allegations of election fraud and promoted “negative narratives” about the US response to the coronavirus pandemic.
Prof Martin Innes, director of Cardiff University’s crime and security institute, said open-source analysis strongly suggested “multiple links” to Beijing.
Researchers initially thought the hidden network was not especially complex, he said. Further evidence, however, revealed what he called a “sophisticated and disciplined” online operation. Accounts did not use certain hashtags in an apparent attempt to avoid Twitter’s counter-measures. They posted during regular Chinese working hours, with gaps on a national holiday, and used machine tools to translate into English.
“The network appears designed to run as a series of almost autonomous ‘cells’, with minimal links connecting them,” Innes said. “This structure is designed to protect the network as a whole if one ‘cell’ is discovered, which suggests a degree of planning and forethought.
“Therefore, this marks the network as a significant attempt to influence the trajectory of US politics by foreign actors.”
Efforts by Russian-linked social media actors to influence US elections are well known. The special counsel Robert Mueller detailed an extensive troll operation run out of building in St Petersburg. Its goal was to “disparage” Hillary Clinton and to promulgate “divisive” content, Mueller found.
The Chinese accounts cannot be definitely linked to the state. But ordinary Chinese citizens do not have access to Twitter and it appears that Beijing may be seeking to emulate Kremlin practices by setting up its own US-facing political influence operation.
Last year the university’s research team uncovered more than 400 accounts engaging in suspicious activities. These were forwarded to Twitter, which suspended them within a few days. The latest analysis suggests further accounts are still working, with the network more resilient than previously thought.
There is compelling evidence of links to China. Posts feature the Chinese language and a focus upon topics reflecting Chinese geopolitical interests. Some 221 accounts spread content in favour of the Chinese Communist party, encompassing some 42,618 tweets, the study found.
The accounts also attacked Trump for referring to Covid-19 as the China virus. One claimed the virus originated outside China and had actually come from the US laboratory at Fort Detrick, in Frederick, Maryland.
The network’s main goal was “encouragement of discord” in the US, the study concluded. Most tweets about Trump were negative. The handful that were positive urged Americans to “fetch their guns”, to “fight for democracy” and to “call gunmen together” in order to win a second Trump term.
The bots complained of “double standards” after the Capitol building riot, saying US politicians had hypocritically backed protesters who entered the Hong Kong legislative building. “The riots in Congress are a disgrace to the United States today, and will soon become the fuse of the American order,” one remarked.