Voters, especially the younger generation who are more digitally connected, are encountering political information in an environment that is shaped by artificial intelligence. From social media platforms to automated news summaries, AI is influencing what kind of information citizens see, how they interpret it, and how they then vote. These systems are expanding in a rapid way across all democracies.
Some questions arose about what kind of information these systems produce, where it goes, and how it works in electoral processes. But the main question is how these systems are both empowering and undermining democratic participation.
AI-Driven Misinformation
The same technologies that can facilitate access to information also can make easier and quicker the production and diffusion of misinformation. Generative AI systems automate and rapidly spread intentional misinformation campaigns. This new way of operating removes the need for human input and more importantly reduces the cost of creating and spreading misinformation. Moreover, misinformation generated by AI systems can sometimes be better written and even more convincing than those made by human propagandists. (Monteith et al., 2023)
This is not just a theoretical concern. A real example of this issue is in the 2024 elections held in Romania. The Romanian presidential election was annulled because of concerns about AI-generated influence campaigns on TikTok. Another example is in Slovakia in 2023, just a couple of days before the parliamentary elections, a deepfake audio started to circulate on social media, appearing to show a liberal party leader discussing how to rig the election. This disinformation spread really fast during the 48-hour pre-elections silence period, when the candidate couldn’t respond publicly.
These examples show how AI tools can be weaponised to overflow social media with a lot of misinformation, specifically during electoral campaigns.
Age-Specific Vulnerabilities to AI Manipulation
The effects of AI are more visible among younger users, who rely more on digital platforms as a principal source for information and news. Research on students that use AI found out that the use of ChatGPT is likely to develop tendencies for procrastination and memory loss and to weaken students’ academic performance. These cognitive patterns reduce critical thinking and engagement with information. These issues have direct implications even for democratic participation. A generation used to receiving algorithmically curated, AI-generated summaries of complex issues may be less prepared to assess critically and autonomously political claims. (Abbas et al., 2024)
A broader analysis of this cognitive erosion is presented in a study from Gerlich’s 2025 study published in Societies. The study surveyed 666 participants across diverse age groups and educational backgrounds. The study revealed a significant negative correlation between frequent AI tool use and critical thinking abilities. This means that the more individuals delegate cognitive tasks to external tools the more they reduce their engagement and reflective thinking. Younger participants are the ones that are at risk because they presented both higher dependence on AI tools and lower critical thinking scores compared to older participants.
For democratic participation this translates in citizens increasingly outsourcing information retrieval, source evaluation and even decision-making to AI systems. For example, instead of comparing information from multiple news sources or deciding which political claim, statement is credible, people are more tempted to use and rely on AI-generated summaries or answers. Same in voting, voters may ask AI systems to explain election manifestos, compare political parties or even recommend which candidate to vote based on their views.
Cognitive offloading is the process of using external tools, technologies, or other people to reduce the mental effort required to remember and process information. Cognitive offloading is strongly correlated with diminished critical thinking, forming a feedback loop: more AI use leads to more offloading, which further erodes individuals analytical capacity.
The study also documented a phenomenon called the “Google effect”. The “Google effect” means that people remember more where to find information rather than the information itself. With AI this suggests that AI tools are reshaping how citizens retain the information. In a political context, this translates in voters that may be less able to recall facts independently, making them more easily influenced by whoever is in control of the information at a given moment. All in all AI doesn’t need to deceive voters directly. It could simply diminish their ability to think independently.
Beyond cognitive impacts, AI content can also narrow the range of perspectives users are exposed to. Algorithms and automated content contribute to a homogenisation of ideas, reinforcing this passive consumption of information rather than active political participation. When the same AI tools shape the information of millions of young voters, the result can be the end of intellectual diversity on which our democracy depends.
Different generations are facing AI-driven electoral risks in different ways, but still, the effects of those risks converge in a stronger erosion of democratic trust. Older populations may be the ones that are believing more in manipulated content such as deepfakes in political videos. However, even younger citizens who have more digital capabilities are not safe. Some research has shown that warnings about deepfakes in political videos may not mitigate the threats. (Aditya Kumar Shukla & Tripathi, 2024)
What is even more worrying is that AI-generated misinformation does not need to be believed to cause harm. Even when false content fails to convince, its quantity and quality create so much confusion that citizens struggle to know what to believe, and they are pushed to simply stop engaging.
Unequal Visibility
These systems reflect and amplify already existing societal inequalities. It has been shown that large gender and racial biases are present in AI systems, which are being sold by major tech companies. When AI systems were asked to identify the gender of a person from a facial image, all of the companies had higher accuracy for male faces than for female faces. Error rates were below 1% for lighter-skinned men but increased to as much as 35% for darker-skinned women. (Buolamwini, 2019)
The underrepresentation of women and people of colour in technology and the undersampling of these groups in AI data have resulted in biased technology. In elections, this creates unequal visibility and unfairly amplifies certain candidates more than others.
The limits of regulation
In response to these challenges, the European Union has introduced measures to address AI-generated misinformation. The Digital Services Act requires very large platforms such as Facebook and TikTok to assess and mitigate systemic risks created by algorithmic content, which may include ensuring deepfakes are clearly marked. The EU AI ACT, whose transparency requirements are set to apply from August 2026, is a more direct approach, requiring creators and users of AI systems to label deepfakes and put watermarks on AI content. Rather than prohibiting harmful deepfakes, the Act prioritises transparency. Nevertheless, this is not enough. Marking AI content is necessary, but it is not sufficient; detecting and removing disinformation generated by AI after it has spread requires stronger approaches.
Technological development continues to surpass EU regulatory capacity and the long-term success of these measures will depend on collaboration between governments, platforms and civil society. (Aditya Kumar Shukla & Tripathi, 2024)
The integration of AI into electoral processes offers useful tools to improve civic engagement and give more access to political information. However, it introduces risks related to misinformation and algorithmic bias that contribute to the decline of institutional trust. Ultimately, the question is not whether AI will shape our democracies, because it already does. The question is whether legislators and developers will act with enough responsibility to ensure it does so fairly. Democratic systems are resilient, but they are not self-correcting.
As Citizens Take Over Europe, we believe that engaged citizens must be at the centre of this conversation, not as passive users of technology, but as active shapers of the rules that govern it.
Sources:
Abbas, M., Jam, F.A. & Khan, T.I. (2024). Is it harmful or helpful? Examining the causes and consequences of generative AI usage among university students. Int J Educ Technol High Educ, 21, 10. https://doi.org/10.1186/s41239-024-00444-7
Buolamwini, J. (2019, February 7). Artificial Intelligence has a problem with gender and racial bias. Here’s how to solve it. Time. https://time.com/5520558/artificial-intelligence-racial-gender-bias/
Gerlich, M. (2025). AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking. Societies, 15(1), 6. https://doi.org/10.3390/soc15010006
Monteith, S., Glenn, T., Geddes, J. R., Whybrow, P. C., Achtyes, E., & Bauer, M. (2024). Artificial intelligence and increasing misinformation. The British Journal of Psychiatry, 224(2), 33-35. doi:10.1192/bjp.2023.136
Shukla, A. K., & Tripathi, S. (2024). AI-generated misinformation in the election year 2024: measures of European Union. Front. Polit. Sci., 6:1451601. doi:10.3389/fpos.2024.1451601
Smith, M. (2025). The Power of AI to Strengthen Civic Engagement. In: Srivastava, B., Nikolich, A., Hickerson, A., Koppel, T. (eds) PROMISE – PROMoting AI’s Safe usage for Elections. Springer, Cham. https://doi.org/10.1007/978-3-031-89853-2_14