A new study from researchers at USC Viterbi’s Information Sciences Institute (ISI) takes a fresh look at previous claims about global information warfare—that different countries are working together to manipulate public opinion across social media platforms.

“Our rigorous statistical analysis demonstrates that no country or state is coordinating with other states,” said Luca Luceri, research assistant professor of computer science at ISI. “These are independent efforts.” 

The findings reassess a 2023 study led by Pennsylvania State University that claimed to have uncovered “intentional, strategic interaction amongst thirteen different states” in online influence operations. The study, published in Nature, suggested coordinated disinformation campaigns between Russia and Iran, and Cuba and Venezuela, a finding that could have implications for global policymakers. 

Where did previous study differ from ISI’s? It comes down to methodology, according to Luceri. There is a critical difference between interaction and coordination. Interaction refers to state-backed accounts engaging with each other through basic platform functions like replies, retweets, or mentions; coordination, in contrast, involves sophisticated manipulation tactics where malicious users strategically flood online channels to create an illusion of public consensus—possibly where none exists. 

The ISI team investigated whether inter-state coordination existed by analyzing multiple online behavioral traces and using both state-of-the-art coordination detection methodologies and control datasets. 

Importantly, the control dataset captured ordinary users from the same regions discussing similar topics during the same time period. Typically, coordination detection methods use computational models to look for deviations from organic user behavior. This baseline allowed researchers to determine whether interaction patterns between state-backed accounts were genuinely suspicious or merely reflected typical social media behavior from users in these countries. 

When the team applied statistical tests to compare behavior from state-backed accounts and organic social media conversation, they found no statistically significant differences, suggesting an absence of coordinated inauthentic behavior. This methodological approach—incorporating a control dataset—had not been utilized in previous research that reported evidence of coordination.

ISI’s findings are not intended to minimize the potential threat of state-sponsored information operations. Rather, they underscore the importance of robust methodologies and control datasets in accurately detecting online coordination. In this case, the key takeaway is distinguishing between basic interaction and strategic coordination. “You can have people from different states talking to each other,” Luceri said. “And this does not necessarily imply coordination.” 

The ISI study will be presented at the 2025 International World Wide Web Conference, held April 28th through May 2nd in Sydney, Australia.

Published on April 15th, 2025

Last updated on April 15th, 2025