Study Examines AI’s Potential Impact on Elections Amid Rising Social Media Misinformation

A simulated terrorist attack at Sydney’s Bondi Beach on December 14, 2025, which left 15 civilians and one gunman dead, quickly became a breeding ground for AI-generated misinformation that spread across social media platforms with alarming speed.

As Australians struggled to process the tragedy, manipulated content began circulating online. A doctored video falsely claimed one of the terrorists was an Indian national. Meanwhile, X (formerly Twitter) was flooded with tributes to a fictional hero named “Edward Crabtree” who never existed. Perhaps most disturbing was a deepfake photo showing human rights lawyer Arsen Ostrovsky—a real survivor of the October 2023 Hamas attack in Israel—portrayed as a crisis actor having fake blood applied by makeup artists.

This pattern of AI-powered disinformation has become increasingly common in crisis situations worldwide. Similar waves of fake content have emerged during events in Venezuela, Gaza, and Ukraine. According to recent data from cybersecurity firm Imperva, approximately half of all online content is now generated or distributed by AI systems.

“Generative AI can create fake online profiles, or bots, which try to legitimize misinformation through realistic-looking social media activity,” explains Dr. Hammond Pearce, Senior Lecturer at UNSW Sydney’s School of Computer Science & Engineering. “The goal is to deceive and confuse people – usually for political and financial reasons.”

To understand the true scope of this threat, researchers at UNSW Sydney created “Capture the Narrative,” the world’s first social media wargame. The experiment challenged 108 teams from 18 Australian universities to build AI bots designed to influence a fictional presidential election, using tactics that mirror real-world manipulation techniques.

The results were striking. During the four-week simulated campaign, more than 60% of all content was generated by competitor bots, surpassing 7 million posts. These bots engaged in sophisticated information warfare, creating compelling but often false content to sway voters toward either “Victor” (a left-leaning candidate) or “Marina” (a right-leaning candidate).

What made the experiment particularly valuable was the use of “simulated citizens” programmed to interact with social media in ways that mimic real-world behaviors. When these digital citizens cast their votes, “Victor” narrowly won. However, when researchers ran the election again without bot interference, “Marina” emerged victorious with a 1.78% swing—proving that the student-created misinformation campaign had successfully changed the election outcome.

“It’s scarily easy to create misinformation, easier than truth. It’s really difficult to distinguish between genuine and manufactured posts,” one competition finalist noted.

The competitors quickly identified effective tactics, including emotional manipulation and negative framing to provoke reactions. Some teams even profiled “undecided voters” for targeted messaging campaigns. Another finalist admitted, “We needed to get a bit more toxic to get engagement.”

Dr. Alexandra Vassar, a Senior Lecturer involved in the research, points out that the simulated platform eventually mirrored real social media environments: “Our platform became a ‘closed loop’ where bots talked to bots to trigger emotional responses from humans, creating a manufactured reality designed to shift votes and drive clicks.”

This manufactured consensus creates what experts call a “liar’s dividend,” where the proliferation of fake content erodes trust in authentic information. Even genuine content faces skepticism, and legitimate critical voices are frequently dismissed as bots or shills, making meaningful debate on complex issues increasingly difficult.

The experiment highlights the urgent need for enhanced digital literacy among the general public. As AI tools become more accessible and sophisticated, the ability to identify and resist manipulation becomes crucial for maintaining the integrity of public discourse and democratic processes.

“What our game shows us is that we urgently need digital literacy to raise awareness of misinformation online so people can recognize when they’re being exposed to fake content,” says Dr. Rahat Masood, another UNSW researcher involved in the project.

The Bondi Beach simulation serves as a sobering reminder that as AI technology advances, so too must our collective ability to navigate an increasingly complex information landscape.