The Coming AI Disinformation Revolution: How Single Actors Could Deploy Swarms to Threaten Democracy
Back in 2016, hundreds of Russian workers commuted daily to a sleek modern office building at 55 Savushkina Street in St. Petersburg. They were staff of the Internet Research Agency (IRA), the now-infamous troll farm that became synonymous with foreign election interference. Working around the clock, seven days a week, these employees manually left comments on news stories, posted content across Facebook and Twitter, and systematically worked to stoke division among American voters ahead of that year’s presidential election.
When the IRA’s operation was finally exposed, it triggered blanket global media coverage, spurred high-profile U.S. Senate hearings, and pushed major social platforms to overhaul their user verification processes. Yet for all the funding and manpower the Russian operation poured into the campaign, its actual impact on the election was negligible—especially when compared to the separate Russia-linked effort that leaked Hillary Clinton’s private emails in the final weeks before voting.
Ten years on, the original IRA is long defunct, but disinformation campaigns have continued to evolve rapidly, with bad actors now leveraging AI to build convincing fake websites and hyper-realistic deepfake videos. A new paper published Thursday in Science forecasts an imminent, transformative shift in how these disinformation operations will be run. Rather than requiring hundreds of employees seated at office desks in St. Petersburg, the paper argues, a single bad actor with access to cutting-edge AI tools will soon be able to command "swarms" of thousands of social media accounts. These swarms will not only craft unique posts indistinguishable from genuine human content, but will also adapt independently and in real time—all without constant human oversight.
These AI swarms, the researchers warn, could shift societal-wide public opinion, sway election outcomes, and ultimately bring about the collapse of democratic governance if preventative steps are not taken immediately.
"Advances in artificial intelligence offer the prospect of manipulating beliefs and behaviors on a population-wide level," the report says. "By adaptively mimicking human social dynamics, they threaten democracy."
The paper was written by 22 international experts, drawn from a wide range of fields including computer science, artificial intelligence, cybersecurity, psychology, computational social science, journalism, and government policy. This grim assessment of AI’s impact on the global information environment is shared by other leading experts who have reviewed the work.
"To target chosen individuals or communities is going to be much easier and powerful," says Lukasz Olejnik, a visiting senior research fellow at King’s College London’s Department of War Studies and author of Propaganda: From Disinformation and Influence to Operations and Information Warfare. "This is an extremely challenging environment for a democratic society. We're in big trouble."
Even experts optimistic about AI’s potential to benefit humanity agree the paper highlights an urgent threat that cannot be ignored.
"AI-enabled influence campaigns are certainly within the current state of advancement of the technology, and as the paper sets out, this also poses significant complexity for governance measures and defense response," says Barry O’Sullivan, a professor at the School of Computer Science and IT at University College Cork.
In recent months, as AI companies race to prove they deserve the hundreds of billions of dollars in investment that has flowed into the sector, many have pointed to the latest generation of autonomous AI agents as proof the technology will finally live up to years of hype. But the paper’s authors argue this very same technology could soon be weaponized to spread disinformation and propaganda at a scale never before seen.
The swarms described by researchers are made up of AI-controlled agents capable of maintaining consistent, long-term online identities and—critically—retaining memory of past interactions, allowing them to simulate believable, authentic-seeming profiles. Agents coordinate with one another to hit shared campaign goals, while building distinct individual personas and generating unique content to avoid detection by platform security systems. These systems can also adapt in real time, responding to algorithmic signals from social platforms and adjusting their approach based on conversations with real human users.
"We are moving into a new phase of informational warfare on social media platforms where technological advancements have made the classic bot approach outdated," says Jonas Kunst, a professor of communication at BI Norwegian Business School and one of the paper’s co-authors.
For experts who have spent years tracking and countering disinformation, the paper outlines a deeply alarming vision of the near future.
"What if AI wasn't just hallucinating information, but thousands of AI chatbots were working together to give the guise of grassroots support where there was none? That's the future this paper imagines—Russian troll farms on steroids," says Nina Jankowicz, the former Biden administration disinformation lead who is now CEO of the American Sunlight Project.
Researchers note it is still unclear whether this tactic is already in widespread use, because current systems built to track and identify coordinated inauthentic behavior are not equipped to detect these AI swarms.
"Because of their elusive features to mimic humans, it's very hard to actually detect them and to assess to what extent they are present," says Kunst. "We lack access to most [social media] platforms because platforms have become increasingly restrictive, so it's difficult to get an insight there. Technically, it's definitely possible. We are pretty sure that it's being tested."
Kunst adds that these systems are still likely relying on some degree of human oversight during development, and predicts that while they may not have a large-scale impact on the 2026 U.S. midterm elections in November, they will almost certainly be deployed to disrupt the 2028 presidential election.
AI-generated accounts indistinguishable from real humans are only one part of the threat. Researchers add that AI’s ability to map entire social networks at scale will let disinformation organizers target AI swarm agents specifically to vulnerable communities, ensuring maximum impact.
"Equipped with such capabilities, swarms can position for maximum impact and tailor messages to the beliefs and cultural cues of each community, enabling more precise targeting than that with previous botnets," they write.
These systems are also inherently self-improving, using responses to their posts as feedback to refine their approach and deliver more effective messaging.
"With sufficient signals, they may run millions of microA/B tests, propagate the winning variants at machine speed, and iterate far faster than humans," the researchers write.
To combat the threat of AI swarms, the research team proposes the creation of an independent AI Influence Observatory, made up of members from academic groups and non-governmental organizations. The body would work to standardize evidence of AI influence campaigns, improve global situational awareness, and enable faster collective responses to threats, rather than relying on top-down punitive measures.
Notably, social media platform executives are intentionally excluded from the proposed observatory, primarily because researchers argue platforms’ business models prioritize user engagement above all else, giving them little incentive to root out these harmful AI swarms.
"Let's say AI swarms become so frequent that you can't trust anybody and people leave the platform," says Kunst. "Of course, then it threatens the model. If they just increase engagement, for a platform it's better to not reveal this, because it seems like there's more engagement, more ads being seen, that would be positive for the valuation of a certain company."
Beyond inaction from platforms, experts add there is little political incentive for governments to step in and address the threat.
"The current geopolitical landscape might not be friendly for ‘Observatories’ essentially monitoring online discussions," Olejnik says. Jankowicz agrees: "What's scariest about this future is that there's very little political will to address the harms AI creates, meaning [AI swarms] may soon be reality."
