The core of the problem is a simple mismatch in speed. The development of AI is happening at a blistering pace, championed by figures from Sam Altman to Elon Musk, while our democratic institutions, by design, are slow, deliberative, and cautious. This creates a gaping vulnerability, a space where technology can run rings around regulation. The real question about AI in democratic processes isn’t whether it will have an impact, but whether our systems can possibly adapt before they are fundamentally broken.
Voter Analytics: A Smart Campaign or a Sinister Script?
Every politician wants to know what voters are thinking. In the past, this meant town halls, door-knocking, and clunky telephone polls. Today, it means voter analytics fuelled by vast datasets and machine learning algorithms. On the surface, this sounds like an upgrade. Campaigns can tailor their message, engage with citizens on issues they genuinely care about, and potentially boost turnout. It’s a personalised political pitch.
But where is the line between personalisation and manipulation? When a campaign knows your browsing history, your online purchases, and the emotional tenor of your social media posts, their messaging can shift from persuasion to exploitation. It’s like sending a salesperson into your home who not only knows you’re worried about your finances but also knows you had a bad day and are susceptible to a comforting lie. This isn’t just a better mousetrap; it’s a form of algorithmic social engineering that can fragment the public square into millions of individual, isolated echo chambers.
The Deepfake Deluge and the Fight for Truth
If voter analytics is the subtle nudge, then AI-generated misinformation is the sledgehammer. We are now firmly in the era of the ‘deepfake’, and it is profoundly corrosive to public trust. Just ask the voters in Slovakia. As documented in a recent Cyprus Mail analysis, a convincing AI-generated audio clip of a liberal party leader seemingly discussing how to rig the election circulated just 48 hours before polls opened in 2023. There was simply no time to effectively debunk it.
This isn’t a one-off. Look at the United States, where a fabricated robocall using an AI-cloned voice of President Biden urged New Hampshire Democrats not to vote in the 2024 primary. Or consider India’s 2024 elections, where AI-generated videos showed candidates appearing to speak fluently in languages they don’t know, blurring the lines of authenticity. The goal isn’t always to make people believe the fake; sometimes, it’s just to muddy the waters so much that nobody knows what to believe at all. This creates what experts call the ‘liar’s dividend’, where even real videos can be dismissed as deepfakes. The response? An arms race, with AI-powered misinformation detection tools desperately trying to catch the fakes that other AIs are creating.
Can We Build a Better Democracy with Code?
It’s not all doom and gloom. Away from the electoral chaos, there are genuinely interesting experiments in using AI to improve how democracy functions. Proponents of new digital governance models envision a future where AI can help manage complex civic tasks. Imagine an AI helping a city council analyse thousands of public comments on a new housing development, identifying common themes and concerns to make better, more responsive decisions.
These models could make citizen participation more direct and efficient. Instead of a single vote every few years, you might have ongoing digital consultations on local budgets or infrastructure projects. The challenge, of course, is ensuring these systems are fair, transparent, and don’t simply create a new form of digital oligarchy, where only those who understand and control the algorithms have a real say. Are we building a more inclusive digital Athens, or are we handing the keys to a new tech-savvy elite?
A Global Phenomenon with Local Consequences
The impact of AI on elections isn’t confined to a few high-profile cases in the West. It’s a global issue manifesting in unique ways. In Latin America, extensive research from 2022 to 2024 has documented waves of coordinated bot activity designed to amplify specific narratives, harass journalists, and create an illusion of grassroots support or opposition. These aren’t sophisticated deepfakes but brute-force digital astroturfing that can poison the online conversation and distort the political landscape.
This shows that the threat isn’t just about high-tech trickery. It’s also about the industrial-scale automation of propaganda. Each country’s information ecosystem provides a different set of vulnerabilities for these tools to exploit, turning elections into global testbeds for new forms of digital influence operations.
Europe’s Answer: The World’s AI Referee?
Faced with this tidal wave, regulators are finally waking up. The European Union has taken the most ambitious step with its Artificial Intelligence Act, which officially came into force in August 2024, and the Digital Services Act. These sprawling pieces of legislation are Europe’s attempt to set the global rulebook for technology.
The AI Act categorises AI systems by risk, placing strict obligations on ‘high-risk’ applications, which includes systems that could “influence the outcome of elections and the voting behaviour.” Essentially, if your AI is powerful enough to swing a vote, Brussels wants to have a very serious word. These regulations, as detailed by the Cyprus Mail, represent a bold attempt to rein in the ‘dangerous master’ without strangling the ‘useful servant’. The big question, however, is whether the EU has the institutional capacity and technical expertise to actually enforce these rules against globally dominant and rapidly innovating tech giants. It’s one thing to write the law; it’s another entirely to police it.
So, where do we go from here? AI is not a monster to be slain, nor is it a saviour. It is a tool, arguably the most powerful one humanity has ever created. The danger lies in our passivity—in letting this technology reshape our civic life without a conscious, public, and deeply democratic debate about the rules of engagement. We must decide whether this tool will be used to enhance our collective voice or to silence it.
What do you think is the biggest AI-related threat to our democratic process right now, and what’s one rule you would impose on its use in politics if you could?
