Colombian presidential candidates face a surge of AI-generated disinformation and deepfakes linking them to armed groups ahead of the May 31 election. These digital fabrications, targeting figures like Iván Cepeda and Paloma Valencia, exploit national sensitivities to conflict to manipulate voter perception through coordinated social media campaigns.
AI-Generated Fabrications and Candidate Deception
The upcoming election to choose a successor to President Gustavo Petro has become a primary target for sophisticated digital manipulation. As the country decides between continuing peace dialogues or implementing strict military crackdowns, bad actors are deploying fake news and AI-generated montages to associate political aspirants with illegal organizations.
Iván Cepeda, who advocates for the government’s "total peace" policy, has been a frequent target of high-tech character assassination.
- An altered video utilizing artificial intelligence that depicts Cepeda claiming he will incorporate guerrillas into the military forces.
- A montage showing the candidate “receiving instructions” from Iván Mordisco, the nation’s most wanted guerrilla leader.
- Disseminated audio of a supposed rebel leader threatening voters in the south of the country if they do not support Cepeda.
The Ministry of Defense recently intervened in the audio controversy, conducting an investigation that revealed the voice belonged to an incarcerated extortionist rather than a rebel commander. This overlap between traditional criminal activity and digital disinformation highlights a growing threat to electoral integrity.
Coordinated Disinformation and the Cost of Anonymity
The technical execution of these campaigns suggests they are not merely the work of spontaneous users. According to the Misión de Observación Electoral (MOE), the digital environment provides a shield for those orchestrating these attacks. Frey Muñoz, the subdirector of the MOE, noted that social media platforms "facilitate anonymity," making it difficult to identify "who is behind the scenes."
Beyond individual actors, the MOE has identified a more organized threat structure. The organization warns of coordinated accounts with clear interests, that put money into a platform to artificially inflate the reach of false narratives. This "pay-to-play" model of disinformation allows fringe or malicious content to bypass organic limitations and reach massive audiences through targeted advertising.
A notable example involved right-wing candidate Paloma Valencia. A video surfaced in which she supposedly "praised" the paramilitary groups of the 1990s; the content was not merely shared but explicitly paged via Facebook advertisements to secure more than 30,000 views. This demonstrates how financial resources are being used to weaponize historical trauma for modern political gain.
Exploiting National Trauma Through Algorithmic Reach
The success of these AI-driven attacks is not solely a technological failure but a psychological one. In a country where armed groups have historically interfered in political processes, the electorate remains deeply vulnerable to narratives involving narcos, paramilitaries, and guerrillas.
“That is useful for the people who spread disinformation.
Rincón explains that Colombians are "more sensitive" to conflict-related news, a vulnerability that digital manipulators exploit by creating high-emotion, high-stakes content. When an AI-generated video or an anonymous audio clip touches on themes of national security or illegal violence, the speed of social media ensures the damage is done long before fact-checkers or government agencies can respond.
As the May 31 election approaches, the battleground has shifted from traditional debate to the metadata of social media. The ability of actors to hide behind anonymity while using paid boosts to spread synthetic media suggests that the next phase of political warfare in Colombia will be defined by the struggle to distinguish human intent from algorithmic manipulation.
