From recommendation systems to election deepfakes, AI is transforming how public opinion is formed and how democratic processes unfold. A journey through risks, opportunities, and tools for protecting civic participation.
Democracy has always thrived on information, debate, and participation. Today, however, these three pillars increasingly pass through algorithmic systems that decide what we see, how we see it, and with whom we discuss it. Artificial intelligence is no longer a simple technical tool: it has become a silent yet powerful actor in the public life of our societies.
The Public Sphere Filtered by Algorithms
Social networks use AI-based recommendation systems to maximize user engagement. This seemingly neutral mechanism has profound consequences for the formation of public opinion: so-called ‘echo chambers’ amplify polarizing content, while moderate or complex news struggles to emerge. The result is often a fragmented public debate, where the quality of information gives way to the ability to capture attention.
Deepfakes and Electoral Disinformation
One of the most discussed risks concerns the use of generative AI to produce fake but extremely realistic content. Synthetic videos, audio, and images can portray politicians saying things they never said, with potentially devastating effects during election periods. Several democracies have already experienced manipulation attempts through deepfakes, prompting governments and platforms to develop countermeasures:
- Digital watermarking systems to identify AI-generated content
- Specific laws requiring disclosure of synthetic content during election campaigns
- Automated fact-checking tools based on language models
- Media literacy programs to help citizens recognize manipulation
AI as a Tool for Civic Participation
Not everything is risk, though. Artificial intelligence also offers unprecedented tools to strengthen democratic participation. Digital deliberation platforms use AI to synthesize thousands of citizen opinions, identifying points of convergence in complex debates such as urban planning or climate policies. Projects like ‘AI-assisted deliberative polling’ allow informed consensus to emerge on controversial topics, moving beyond the yes/no logic typical of traditional referendums.
Some local governments are experimenting with institutional chatbots that help citizens understand complex laws or participate in public consultations, breaking down linguistic and bureaucratic barriers that traditionally exclude the most vulnerable segments of the population.
Algorithmic Surveillance and Its Limits
A delicate chapter concerns the use of AI by states for social control: facial recognition, predictive behavioral analysis, social media monitoring. While in some contexts these tools are presented as public safety measures, in others they become mechanisms of dissent repression. The line between security and surveillance is increasingly thin, requiring solid regulatory frameworks like the European AI Act, which classifies some AI uses as ‘unacceptable risk.’
Toward Democratic Governance of AI
The main challenge for contemporary societies is not so much stopping AI development, but governing it according to democratic principles: algorithmic transparency, platform accountability, citizens’ right to understand how decisions affecting them are made. Only through participatory and informed governance will it be possible to harness AI’s potential to strengthen, rather than weaken, the foundations of democracy.
The future of the public sphere will depend on the collective ability to steer these tools toward the common good, keeping the irreplaceable value of informed and conscious human participation at the center.