Colombian military personnel installing an ART SCANEAGLE-NIGHTEAGLE drone during the International Aeronautics Fair—Colombia 2019 at the José María Córdova International Airport in Rionegro, Colombia, on 11 July 2019. Image by Staff Sgt. Megan Floyd, from the Flickr page of the S.C.  Air National Guard. Public domain.

This article is part of Global Voices’ special series for April 2026, “Human Perspectives on AI”. This series will offer an insight into how AI is being used in countries of the global majority, how its use and implementation are affecting specific communities, what this experimenting with AI could mean for future generations, and more. You can support this coverage by making a donation here.

Artificial intelligence (AI) does not act neutrally in contexts of war. In Colombia, it is gradually gaining importance, amplifying inequalities, accelerating disinformation, and reconfiguring violence.

Through five cases reported by local media, this article explores how accessible technologies are reconfiguring different dimensions of Colombia's armed conflict which, despite several peace processes, has persisted for over 60 years due to the presence of multiple non-state armed groups, illegal economies, and deep regional inequalities. More than simply a uniform adoption of artificial intelligence, a technological innovation has emerged that is transforming not only the battlefield, but also decision-making processes, local narratives, and forms of social control.

Low-cost drones

In several regions of Colombia, since 2024, non-state armed groups have been using modified commercial drones in rudimentary ways to attack police stations and military positions with explosives, leaving hundreds of uniformed personnel injured or killed. While these devices do not incorporate AI, their use reflects the ability of armed actors with limited resources to adapt accessible technologies and thereby change the tactical balance of power. 

In more sophisticated and scale-produced systems, as seen in the war in Ukraine and in conflicts linked to Iran, drones are part of more complex military strategies. In the Colombian context, though, use of drones represents an intermediate stage — a war still conducted by humans, but approaching forms of automation that could integrate more sophisticated algorithmic tools in the future.

More sophisticated anti-drone systems

The transformation of the battlefield with the use of low-cost drones has forced a technological response from the state.

Since 2025, the Colombian government has been creating a “national anti-drone shield” — a complex technological architecture that combines multiple capacities: specialized sensors capable of detecting drones and differentiating them from birds or aircraft; technologies such as micro-Doppler to identify moving objects; radio frequency systems that locate and block the signal between the drone and its operator; and both electronic and physical neutralizating mechanisms.

This hybrid platform, with its combination of physical sensors, data processing algorithms, and human decision-making, is gradually redefining how threats are detected, evaluated, and responded to.

Algorithms that map risks and operations

Security institutions have started to incorporate AI tools to guide operational action, based on data analysis that allows them to identify risks and anticipate threats. The Police Service Model, adopted in 2024, proposes a “focused and differentiated” deployment based on real-time data analysis to identify crime hotspots. 

This approach has also been adopted by the military. Reports by the Colombian Aerospace Force point to the use of surveillance and reconnaissance systems that integrate advanced sensors and data processing to build risk models in conflict zones, facilitating the most accurate deployment of capabilities on the ground and in the air.

These systems do not operate autonomously; rather, they structure the map on which action is taken. In a context characterized by regional inequalities and limited data quality, the promise of greater efficiency in security comes with the risk of reproducing existing biases and intensifying surveillance in stigmatized areas, while others remain invisible.

Manipulating narratives

In 2023, local media documented the increasing use of AI to mislead people in sensitive political contexts. During the regional elections in October, which saw local power disputes in areas affected by armed groups, artificially generated audio messages began to be shared using the voices of election candidates, to attack or benefit their campaigns. 

That November, videos with AI-generated avatars pretending to be doctors, long-term patients, and military personnel calling for protests against the government, were shared online. This sophisticated production of fake voices and images can alter public perception and influence security decisions, stigmatize communities, and legitimize measures of control.

The social basis of algorithmic control

Community groups on social media that work — in theory — to share security alerts, have been co-opted by non-state armed groups to share photographs and profiles of community leaders, accusing them of collaborating with rival groups or with the government.

This tends to happen mainly in closed WhatsApp groups created by residents under pressure from armed actors, where intimidating messages and unsubstantiated accusations are being shared. On Facebook, anonymous or fake profiles amplify this content. These digital blacklists, linked to threats, forced displacement and targeted killings, are the basis of a system of control that could be accelerated and automated by AI.

These cases demonstrate how AI is gradually being introduced into the Colombian armed conflict, where accessible technologies, data analysis, and partial automation reinforce and amplify pre-existing structures of violence. They make surveillance more efficient, disinformation faster, and state intervention potentially more unequal.

In this context, the risk is not only that the conflict evolves into a war dominated by autonomous systems; the most immediate danger is that these tools end up reinforcing the same regional and social disparities that have fuelled violence for decades.