From Colombia to Peru, Brazil, the United States and Spain, political and electoral dynamics are being reshaped by the impact of artificial intelligence. A wide range of AI tools is being used at every stage of the electoral cycle and by all the actors involved—not only in campaign communications and strategies, but also in how people obtain information and participate in politics, how they decide whom to vote for, and how they interact with political actors and state institutions. Somewhat further behind, political parties, candidates and state institutions are seeking to transform their own practices and offerings in response to this new reality.
Just two years ago, much of the global conversation about artificial intelligence and elections was marked by catastrophic predictions. There was talk of a flood of malicious uses of AI to alter results, manipulate public debate and render the state institutions responsible for organizing elections irrelevant. But the global experience of the 2024 and 2025 electoral supercycle, as well as the most recent elections in 2026, forces us to take a somewhat more nuanced view. Less alarmist, but inexorably much more proactive about the future.
A report by the German Marshall Fund, consistent with other research, detected significant deepfake campaigns in 39% of the elections monitored since late 2023 across 30 countries.
The 2026 Latin American elections have been no exception to this trend. On the contrary, they have shown that fraud narratives have been fueled by massive disinformation facilitated by AI. Thus, the risks have not disappeared; they continue to grow and mutate as we confront them.
What we are observing today is not a snapshot, but a fast-forwarded film. An increasingly diverse range of uses for new technologies. Artificial intelligence is being used by everyone, but also by malicious actors seeking to subvert the popular will deliberately.
In this new scenario, we already know many of the emerging risks: algorithmic discrimination, disinformation, synthetic content, violations of data privacy, the influx of illicit financing into politics, and the manipulation of public debate. AI is also amplifying pre-existing structural inequalities, such as political violence against women.
Faced with this new reality, democratic institutions are not mere spectators. A study by International IDEA, Artificial Intelligence and Electoral Management: A Map of Latin America and the Caribbean, confirms that electoral institutions in Latin America are moving from a stage of exploration toward one of adopting new technological tools and digital governance. Electoral institutions are already using or evaluating AI-based applications to automate administrative tasks, assist with citizen services, support institutional communications, strengthen cybersecurity, and improve processes for monitoring electoral spending, as well as other internal processes involving analysis and planning. However, the report also warns of significant capacity gaps and, in most countries, the absence of specific regulatory frameworks, as well as the need to develop principles of transparency, human oversight, data protection and accountability that enable the responsible incorporation of these technologies.
That is why the question is no longer whether these risks will materialize. The relevant questions are not when they will occur, but at what scale and speed their uses will evolve, what new risks will emerge, and whether our institutions will have the capacity to anticipate and respond to them.
Doing nothing to adapt and protect elections, or simply waiting for a crisis to occur before taking action, is neither a possible nor a responsible alternative. Elections are, no more and no less, the pillar that protects the reproduction and survival of democratic regimes.
For decades, we thought that electoral trust depended primarily on the technical capacity to organize transparent and efficient elections. That dimension remains indispensable, but today it is no longer sufficient. Trust in elections also depends on the quality of the information environment and on the capacity to respond to the increasingly widespread narratives of “electoral denialism.”
There is no doubt that we do not know all the ways in which AI may impact democracies. But if anything is clear, it is that no institution or country can face these new and constantly changing challenges on its own. Building democratic governance of these technologies requires a genuine democratic ecosystem in which public institutions, technology companies, academia, research centers, civil society and international organizations share knowledge, develop standards and assume common responsibilities. It also requires a transnational alliance to confront the monopolistic power of the major technology companies.
The future of democracy is at stake, and the history of AI is not predetermined. It depends on human agency. Hence the need to continue working urgently and with an eye to the future to protect the integrity of elections as the pillar that sustains and safeguards the sustainability of our democracies.
This text was written in connection with the launch of the report Artificial Intelligence and Electoral Management: A Map of Latin America and the Caribbean, published by International IDEA.
Autor
Regional Director for Latin America and the Caribbean at the International Institute for Democracy and Electoral Assistance (International IDEA). PhD in Political Science from the University of Wisconsin-Madison, United States. Former Minister of Justice and Human Rights of Chile.
