Democracy Being Hijacked Across Nations

Beyond the benefits of Artificial Intelligence (AI), alarming data has recently got surfaced. A new study reveals that AI can mimic human behavior almost perfectly on the internet and social media, intentionally spreading misinformation to manipulate elections and engineer consensus by influencing others.

A study published in the journal Science has labeled this state-shattering AI activity as "Democracy Hijacking." Between 2017 and 2024, AI has reportedly interfered in the elections of 70 countries, ranging from the Philippines to the United States. Similar successes in spreading fake information were observed in Brazil and Ireland, too.

One of AI's most significant capabilities is its ability to operate across multiple countries simultaneously. It deploys a multi-agent framework with full force to overturn the political landscape of targeted nations. Under a Large Language Model (LLM) system network, a "central controller" sends instructions to LLM agents, which experts in the study refer to as "fake citizens."

Each agent works independently, filtering public posts and mimicking specific writing styles. They then precisely spread fabricated viewpoints across social media tailored to the situation. Following this, these agents rapidly plot their next moves, creating a false image of democratic consensus. This "AI swarm" of autonomous LLM agents automatically distorts political discourse, joining online communities to create a facade of massive support for specific political positions.

The scope of these attacks is not limited to social media. AI easily infiltrates public opinion polls, online petition platforms, local government feedback portals, and community forums, creating the illusion of widespread grassroots support for even the most obscure issues.

How it Impacts Democracy

Deepfakes: By mimicking the voices or images of political leaders, AI creates fake audio and video (deepfakes) to mislead voters. For example, during the U.S. elections, robocalls using a cloned version of Joe Biden's voice were sent to discourage people from voting.

Targeted Campaigns: AI is used to analyze voter preferences. Highly specific political messages or advertisements are then delivered to individuals, proving effective in shifting public opinion.

Preventative Measures

International organizations are emphasizing the need to ensure this technology strengthens democracy rather than undermining it. Bangladesh is also considering the formation of a coordinated cell to combat misinformation and the misuse of AI in elections.