Ultimately, the answer to AI and fake news will not come from one measure or one institution. It will be a combination of regulation, technological solutions, professional standards in the media and, perhaps most importantly, long-term investment in media and digital literacy.
The development of generative artificial intelligence in the last few years has radically changed the way political messages are created, public opinion is shaped, and electoral campaigns are conducted. AI has become a tool that facilitates research, analysis, and communication, but at the same time, it is the most dangerous instrument for manipulating voters.
In political campaigns, where emotion and speed often override facts, AI opens up space for sophisticated deceptions that can influence the election outcome and undermine trust in democratic processes in the long term.
Generative tools like Midjourney allow for the creation of convincing photos, videos, and audio recordings in minutes. Experts warn that we are entering an era in which “evidence” will become the most vulnerable category of public life.
Since 2021, deepfake videos have become more available and convincing due to the vast amount of data fed into AI image generators. This means that AI-generated content is increasingly present in politics and election campaigns, sometimes sowing confusion among voters and causing them to lose trust in the veracity of what is being presented to them.
“This can lead to misinformation, people can be confused by what they see,” says Lisa M. Given, professor of information sciences at RMIT University in Melbourne.
The technology is being used as a new “low- or medium-effort political attack meme” to defame or ridicule opponents, according to Scott Steinhardt, director of marketing at deepfake detection company Reality Defender.
“Fake images and other forms of disinformation are mostly spread via social networks, although traditional media such as television and radio can also be distorted”, says Adeboye Adegoke, senior manager at digital rights organization Paradigm Initiative.
Danielle Citron, a law professor and one of the leading authors on the subject of digital manipulation, warns of the phenomenon called “liar’s dividend”. “The real threat of deepfakes is not just that people will believe what is false, but that they will stop believing what is true.”
Hany Farid, a professor of digital forensics at Berkeley, warns in similar fashion. He points out that deepfakes are already so convincing that the average citizen has no chance of discerning them.
Presidential elections in Argentina in 2023 became a global example of how AI can shape political perception. Both sides used generative tools: Javier Milei posted a fake image of Sergio Massa as a communist soldier, while Massa’s team responded with AI-generated depictions of Milei as a zombie and a pirate. This case showed how AI can become an integral part of political strategy, rather than just an isolated incident.
In the last presidential election in Turkey, the main opponent of President Recep Tayyip Erdogan, Kemal Kilicdaroglu, accused Russia of publishing deepfake content in an attempt to influence the result.
Similar cases were witnessed in several European countries, but most cases in Europe have, so far, included “critical art, satire, memes and misuse of images against female politicians”, said Henry Ajder, an expert on deepfakes and generative artificial intelligence.
The Balkan context: fertile ground for manipulation
Miloš Đajić from the Belgrade Center for Modern Skills, one of the leading experts on political campaigns in the Balkans, tells ResPublika that AI is not a fundamentally new threat, but a continuation of existing problems in information. He says that the greatest danger lies not in the technology itself, but in its misuse.
“Artificial intelligence will pose no greater danger to elections than it already poses a danger in information and communication. The greatest danger is the misuse of photos and the creation of videos with the candidate’s voice that is artificially generated”.
Đajić emphasizes that the key is in regulation and prevention, reminding that many countries have not yet adequately regulated social networks, and are already facing a new generation of digital threats. According to him, voter education and abuse reporting systems must be a priority.
The region has been exposed to disinformation, bot networks and political manipulation for years. AI only increases the capacity and speed of these operations.
Emir Zulejhić from Raskrinkavanje warns that the manipulation of visual content is already standard part of campaigns: “What was Photoshop yesterday will be a deepfake video tomorrow.”
Croatian data analyst Marko Rakar warns that AI will enable hyper-personalized political messages which will be invisible to everyone except the target person. Such fragmentation of information, he says, breaks up the common public space and creates parallel political realities.
Political scientist Vuk Vučetić points out that AI disinformation has a particularly strong effect in societies with low trust in institutions, where citizens often accept emotionally charged content without critical reflection.
Lejla Turčilo, a professor at the Faculty of Political Sciences in Sarajevo, warns that citizens are already overwhelmed with information, and AI further obscures the line between truth and manipulation, which makes it difficult to make informed decisions.
Miloš Vukanović from the Center for Democratic Transition (CDT) from Podgorica believes that AI will certainly be used in campaigns, but that the key question is who will be the first to cross the red line and use deepfake to directly discredit the opponent.
Slovenian analyst Lenart Kučić warns that smaller countries are particularly vulnerable because information spreads quickly and with no criticism, and the media ecosystem is smaller and easier to destabilize, while digital forensics expert from Bulgaria Petar Markov believes that the Balkans is an “ideal environment for AI manipulation” due to political polarization and poor media literacy.
Citizens between overwhelming and apathy
Research shows that exposure to deepfake content does not necessarily change one’s vote, but it can reduce trust in institutions and the media. This leads to digital nihilism, that is, the belief that “everyone lies” and that “nothing is safe”, which is perhaps the greatest danger for democracy.
The common denominator of all these warnings is clear; AI does not create new problems, but radically amplifies existing weaknesses in Balkan societies. In an environment where the media is under pressure, citizens are overloaded with information and political actors are prone to manipulation, AI becomes a risk multiplier.
If we ask the question how prepared the media, citizens and political parties in the Balkans are for campaigns in which AI-generated content can dominate, the answer is: partially aware, but insufficiently prepared.
- Media have a key role to play, but lack the resources, knowledge and often the political independence to play that role consistently.
- Citizens intuitively feel that “something is wrong” with the information space, but without systematic education they are left to their own improvisations.
- Political parties largely see AI as another tool in the struggle for power, rather than as a risk to the very infrastructure of democracy.
Ultimately, the answer to AI and fake news will not come from one measure or one institution. It will be a combination of regulation, technological solutions, professional standards in the media and, perhaps most importantly, long-term investment in media and digital literacy.
If there are no systemic education, regulations and technological solutions, AI will not only change campaigns, but will also change the very idea that elections can be based on informed choice and not on an algorithmically designed illusion.
