In 2017, Samuel Woolley and Douglas Guilbeault—experts in computational propaganda—investigated the role of political bots in the 2016 U.S. presidential election. They examined how automated accounts could enter political discussions, amplify a politician’s messages, and create the appearance of widespread political support. Ten years later, the recent development of LLMs (Large Language Models), in addition to their growing commercial accessibility, has changed the literature of computational propaganda from asking “How many accounts are repeating this message and what are the effects this has on people” to “How much original persuasive material can a single actor produce, and how well can tools and people distinguish it from the real thing?” In this article we will examine computational propaganda and its evolution.
Consensus Gentium is a latin phrase meaning “agreement with the people”; it is often associated with philosophers such as Cicero who used it to prove the existence of the divine. In its essence, Consensus Gentium is the thought that a belief widely or universally held by humans carries a presumption of truth; however, despite modern philosophers critiquing it as a fallacy, there remains a certain validity to it. The core of the Consensus Gentium comes from the way in which people are more likely to believe a statement that is universally accepted—which is also called the bandwagon effect. Yet, what happens when this consent is faked by others? In the context of AI bots, this is referred to as manufactured consensus. Historically, propaganda has been created for this very purpose of constructing consensus, and in our current age of technology, the propaganda has been adapted to be used in social media platforms via political bots. In the 2017 study, Woolley found that bots were capable of influencing political processes of global significance, such as the 2016 U.S. election. According to Woolley, “These results indicate that bots infiltrated the core discussion network of our sample, suggesting that they had the capacity to influence political discourse over Twitter,” demonstrating how as of 2016, simple political bots with capabilities of only sending and resending preconstructed messages were able to influence people’s opinions. Overall, Woolley analyzed how political bots were able to influence politics by manufacturing consensus.
However, modern political bots are more complex than the bots Woolley had been investigating;unlike them, current LLMs are able to reproduce false visual imagery on a scale that wasn’t accessible at the time. According to the Brennan Center for Justice’s 2026 audit, publicly available tools like ChatGPT, Gemini, Grok, Claude, Perplexity, DeepSeek, plus Runway and Flux.2 generated misleading electoral images and videos when asked to. This raises a new question, different from the one before: while past political bots created truth from false consensus, current political bots generate truth through fake evidence.
In conclusion, computational propaganda has evolved with the technology used to produce it. Earlier bots relied on automation and repetition to manufacture the idea of consensus. Meanwhile, current bots make use of larger processing capabilities that allows them to generate false information to fabricate truth. This does not mean constructed consensus has disappeared, but rather that it can now be combined alongside convincing artificial content. Therefore, computational propaganda is no longer just a question of how many automated accounts a person is able to generate, but also how convincing their generated content is.
Bibliography:
Woolley, S. C., & Guilbeault, D. R. (2017). Computational Propaganda in the United States of America: Manufacturing Consensus Online (Working Paper 2017.5). Oxford Internet Institute, Computational Propaganda Project.
Woolley, Samuel C., and Philip N. Howard. Computational Propaganda: Political Parties, Politicians, and Political Manipulation on Social Media. Oxford University Press, 2019.
Brennan Center for Justice. (2026, August 11). Does AI Fight or Fuel Election Disinformation?https://www.brennancenter.org/our-work/research-reports/does-ai-fight-or-fuel-election-disinformation
(MLA citations)