How AI Influences Election Campaigns: Examples of Disinformation
Artificial intelligence is revolutionizing how we create and consume content, but this innovation comes with significant risks, especially regarding elections. AI's ability to generate convincing yet false information poses a serious threat to electoral integrity and democratic processes. This article will explore how AI can influence an election campaign in Romania and examine instances of AI-driven disinformation in other countries.
One of the most challenging tasks for election campaign communication teams has been voter profiling. With databases, AI can now create detailed voter profiles based on demographics, social media activity, and other public sources. Understanding voter types helps campaigns tailor messages and target specific groups for maximum impact. AI can gather information to profile voters effectively and categorize them, ensuring that the campaign message resonates with each group. AI can personalize content based on user profiles, leading to "information bubbles." Voters may be repeatedly exposed to information that confirms their existing beliefs, limiting exposure to diverse viewpoints. This over-personalization can polarize society and influence electoral decisions, potentially undermining a healthy democracy.
Furthermore, AI optimizes advertising campaigns by automatically adjusting content and frequency based on performance. Ads are adapted to maximize reach and engagement with specific target groups, reducing costs and increasing impact. Additionally, candidates now utilize chatbots to interact with voters, answer questions, and initiate conversations by sending messages with campaign content.
By 2024, AI has advanced to analyze voter sentiment and monitor public opinion, gauging public perceptions of candidates through social media data. This analysis allows campaigns to adapt strategies based on real-time feedback and public reactions to different messages.
Amplifying Disinformation
However, as in any political contest, not everyone plays fair. Various tricks and tactics are employed, reflecting a mindset where winning is everything, regardless of the methods used. But the methods do matter.
AI tools like ChatGPT and DALL-E can generate high-quality fake content, including text, images, and videos. This capability can be exploited to create and spread disinformation, making it difficult for voters to distinguish between real and fabricated information. The potential impact on voter opinions and election results is alarming. Political actors, including governments, now utilize AI to manipulate public sentiment. Deepfake videos mimicking political figures and AI-generated propaganda can shape public perception and erode trust in democratic institutions, undermining the integrity of the electoral process.
AI-generated disinformation is often sophisticated and challenging to detect. Current AI models prioritize generating plausible content over verifying its accuracy. This makes AI-generated disinformation harder for fact-checkers and detection systems to identify, allowing false narratives to blend with real ones.
As mentioned earlier, AI models are often trained on publicly available data, with many reputable sources restricting access to their content. Consequently, AI relies on less credible sources, increasing the likelihood of generating biased or false content, negatively impacting public opinion and the electoral process. AI tools enable the automation of disinformation campaigns at an unprecedented scale. Automatically generating and distributing false content on social media can rapidly and broadly influence voter perceptions, making combating disinformation even more challenging.
Who is responsible for this? The lack of clarity regarding accountability for AI-generated content poses a significant challenge. This ambiguity makes it difficult to hold those who misuse AI to undermine electoral integrity accountable, exacerbating regulatory and law enforcement issues. Amplifying disinformation through AI tools like ChatGPT and DALL-E, which can produce high-quality fake content, including text, images, and videos, raises concerns. This capability can be exploited to create and spread disinformation, making it increasingly difficult for voters to differentiate between real and fabricated information. The potential impact on voter opinions and election results is alarming. It only takes one party to utilize AI as a weapon for others to follow suit.
Manipulating Perceptions Through "Deepfakes"
Deepfake technology creates highly realistic video or audio content that can appear authentic. This can be used to distort reality, discredit candidates, or manipulate public opinion convincingly, affecting voting decisions and the integrity of the electoral process.
Furthermore, the continuous spread of disinformation and false messages through AI can discourage citizens from participating in elections. Distrust in candidates and confusion about the electoral process can lead to voter apathy, negatively impacting the representativeness and legitimacy of the democratic process. The spread of AI-generated disinformation creates a climate of distrust. Bombarding voters with conflicting or false narratives can erode trust in the electoral system and reliable information sources. This erosion of trust poses one of the most significant long-term risks to democratic institutions.
AI's ability to mimic candidates, fabricate events, and spread polarizing messages can destabilize elections. Misleading voters can lead to distorted opinions or even electoral apathy, affecting voter turnout and weakening democratic engagement.
Solutions to Combat Deepfakes
To address these challenges, regulatory frameworks must be updated to include guidelines on using AI in public and political contexts. These rules should focus on transparency, accountability, and preventing AI misuse in elections. Developing advanced detection tools is crucial to identify and combat AI-generated disinformation in real-time. Machine learning models and AI systems that focus on identifying synthetic content can help fact-checkers and platforms address disinformation effectively. Increasing public awareness of AI's capabilities and risks is essential. Educating the public about identifying AI-generated content can make them more resilient to disinformation campaigns.
The intersection of AI and elections is where technology meets democracy. AI's potential is immense, but so are the risks if left unchecked. We must take proactive steps through regulations, detection technologies, and education to ensure AI strengthens, rather than undermines, our democratic processes.
Examples of Real-World Disinformation:
- Estonia: During elections, a false claim spread on social media suggesting voters should bring their own pens to polling stations, as those provided were "erasable" and could alter votes.
- Turkey: A disinformation story claimed a German deputy declared that a mosque in Istanbul would be demolished if President Erdogan wasn't re-elected, aiming to erode trust in the electoral process.
- France: Before the presidential elections, extremist groups spread disinformation against COVID-19 vaccination and restrictions, promoting conspiracy theories to create distrust in institutions.
- Germany: Far-right groups propagated narratives about electoral fraud, mirroring election denial movements in the US, to undermine democratic processes.
- Moldova: President Maia Sandu was targeted with deepfakes portraying her supporting a pro-Russian party to undermine her pro-Western government.
- Slovakia: Disinformation campaigns included AI-generated audio clips of political leaders seemingly discussing election manipulation, widely shared before being debunked.
- Spain: False stories about foreign interference and electoral fraud circulated in the 2023 elections, including claims that EU officials campaigned for specific candidates.
- Brexit: AI-generated articles and posts were used to mislead the public about the consequences of Brexit, contributing to voter polarization.
You can find the original article here
Alexandru Dan
CEO, TVL Tech
