Study: AI can already run disinformation campaigns on its own

Artificial intelligence agents are already capable of planning and coordinating disinformation campaigns across social media platforms without human direction, according to a new study cited by a report from the New York Post.

The research, conducted by USC’s Information Sciences Institute and titled “Emergent Coordinated Behaviors in Networked LLM Agents,” found that networks of AI agents could plan, coordinate and execute simulated influence operations on a social media environment modeled after X, amplifying one another’s messages once a human operator sets an initial goal. “Our paper shows that this is not a future threat: It’s already technically possible,” said Luca Luceri, the study’s lead scientist, according to the report.

Unlike older-generation bots, which merely amplify content defined in advance, Luceri said “generative agents are now capable of organizing influence campaigns in a fully automated way and creating credible content that can resonate with certain demographics.” Researchers warned such coordinated networks could flood platforms with manufactured consensus, making fringe views appear mainstream and further eroding public trust — with implications reaching beyond elections into public health, immigration and economic policy debates.

The study, accepted for publication at The Web Conference 2026, notes that a human is still required to set the campaign’s initial goal, though no further guidance is needed once the AI agents begin operating.