In summary:
A social media bot network linked to a Chinese university and an AI company was uncovered on X.
The information warfare network called “Green Cicada” is one of the largest known networks.
What happens next?
There are concerns that the “Green Cicada” operation was staged to disrupt the upcoming US presidential election.
A network of at least 5,000 AI-controlled accounts has been uncovered as part of an alleged Chinese-led information warfare campaign to spread divisive political discourse on social media platform X.
Local cybersecurity firm CyberCX says it has uncovered an operation linked to a Chinese university and an AI company that appears to primarily target controversial American narratives, but has also sometimes addressed Australian content.
“Although the information operations capability is currently relatively ineffective, we believe it could be abused for malicious activities in the future,” the company warns in a recently completed report.
Researchers believe that the cluster of at least 5,000 inauthentic X-accounts, dubbed the Green Cicada Network, is almost certainly controlled and coordinated by an artificial intelligence-based Large Language Model (LLM) system.
The suspected mastermind of this operation was identified as an employee of a Beijing-based AI company who studied at Tsinghua University, which has close ties to the People’s Liberation Army and Beijing’s intelligence apparatus.
“The network is increasingly participating in political discourse, but most accounts remain inactive,” says the report, which has already been forwarded to various federal agencies.
Green Cicada accounts could be staged to disrupt US elections
While the Green Cicada Network primarily focuses on political and cultural issues in the United States, it has also been observed amplifying controversial political issues in Australia, the United Kingdom, Western Europe, India, Japan, and other democratic countries.
“We have observed limited amplification of Australia-specific topics and posts from purportedly Australian individuals. Amplifying topics include support or opposition to political candidates, nuclear energy, the economy, housing, migration, protests, and foreign policy,” CyberCX noted.
“Here we have a false network infiltrating our democratic discourse and not necessarily trying to support one side or the other of these debates, but trying to drive a wedge between the sides of this debate and trying to deepen the division and polarization,” said spokeswoman Katherine Mansted.
According to CyberCX, the network “could potentially be used to interfere in the upcoming presidential election.” The company says it has observed “operational execution improving over time, with activity increasing sharply since July 2024.”
The cluster of up to 8,000 inauthentic accounts is considered one of the largest to be publicly disclosed to date and may be the first significant China-related information operation to have generative AI at the heart of its activities.
“We believe that if all available accounts were used, they could successfully spread polarizing content to sow discord and undermine trust in civil institutions,” the report concluded.
A number of the accounts were identified by using instant injections to force malfunctions that override the original instructions of an AI model.
When analyzing a group of Green Cicada accounts, it turned out that the account they interacted with most often was that of the owner of X, Elon Musk.
Last month, Australia and key partners in the region accused a Chinese intelligence agency of cyber espionage, saying it was a large-scale operation that stole hundreds of usernames and passwords and targeted government and corporate networks.
