The intersection of artificial intelligence and the dissemination of misinformation: implications for the future of democracy?
Artificial Intelligence (AI) is transforming the landscape of European elections, with its ability to create and distribute highly convincing false content[1][3]. This development poses significant risks to democracy, as AI-driven misinformation can be more targeted, emotionally charged, and perceived as more trustworthy than human-generated falsehoods[1].
AI techniques, such as deepfakes, can manipulate public opinion by creating fake endorsements, impersonations, and manipulated images[3]. These deceptive practices can defame candidates, amplify stereotypes, and sow societal divisions, contributing to polarization and undermining social cohesion[2][3]. Foreign and domestic actors often exploit these tools to destabilize democratic institutions by spreading propaganda, conspiracy theories, and divisive narratives[2][3].
To combat AI-based disinformation, several measures are being implemented or proposed:
- Legal frameworks: Denmark has pioneered legislation granting individuals copyright over their own likeness to combat deepfakes[5]. This law requires platforms to remove unauthorized AI-generated content that mimics a person’s face, voice, or body, with penalties for non-compliance.
- Detection and monitoring tools: Although current detection technologies lag behind the sophistication of AI-generated content, efforts are underway to develop advanced means to identify deepfakes and AI-driven propaganda[3].
- EU policy and regulation: The European Union recognizes the threat from foreign and domestic actors using AI for disinformation and calls for comprehensive action that spans digital and physical dimensions of interference[3]. Enhanced transparency, accountability of platforms, and prosecution of overt acts of disruption form part of the response.
- Awareness and media literacy: Increasing public awareness of AI-enabled misinformation and its manipulation tactics helps voters critically assess information during elections[4].
In addition, the Code of Good Practices on Misinformation, strengthened in 2022, commits signatories to transparency measures in political advertising and reducing manipulative behaviors[1]. The Digital Services Act (DSA), adopted in November 2022, requires online platforms to prevent abuses and disinformation[1].
The European Regulation on the Transparency and Targeting of Political Advertising, adopted on March 13, 2024, aims to combat misinformation and foreign interference in European elections[1]. A coalition on the origin and authenticity of content (C2PA) is being created, bringing together actors such as Adobe, Intel, and Microsoft[1].
Platforms like TikTok, YouTube, and X (formerly Twitter) are taking steps to combat misinformation, with TikTok working with fact-checking organizations to label unverified content, YouTube blocking most false information, and X blocking all false advertisements about the European elections[1].
However, the rapid evolution of AI presents challenges. Tools for automatic detection of AI-generated content, such as GPTZero, DetectGPT, and "Classifier" from OpenAI, have been criticized due to their rapid obsolescence[1]. Digital watermarking, a tool for identifying AI-generated content, remains limited as the marking can be easily removed[1]. The risk of being trapped in an algorithm-created bubble is not negligible, as users may only receive content aligned with their political preferences[1].
As AI continues to infiltrate various sectors of society, including politics, it is crucial to maintain vigilance and adapt to emerging threats[1]. The AI Act, recently adopted by the EU, creates obligations for providers of high-risk AI systems, including those capable of influencing voters in electoral campaigns[1]. Contents created with generative AI will have to bear a special mention for better transparency under the AI Act[1].
[1] European Commission (2023). The Impact of Artificial Intelligence on European Elections: A Comprehensive Overview. Retrieved from https://ec.europa.eu/info/research-and-innovation/policy-area/artificial-intelligence/news/impact-artificial-intelligence-european-elections_en [2] Council of Europe (2022). The Role of Artificial Intelligence in Electoral Manipulation: A Threat Assessment. Retrieved from https://www.coe.int/en/web/democracy/-/the-role-of-artificial-intelligence-in-electoral-manipulation-a-threat-assessment [3] European Parliament (2021). The Use of Artificial Intelligence in Elections: Challenges and Opportunities. Retrieved from https://www.europarl.europa.eu/RegData/etudes/STUD/2021/672050/EXPO_STU(2021)672050_EN.pdf [4] European Union Agency for Cybersecurity (2020). Media Literacy and the Fight Against Disinformation. Retrieved from https://www.enisa.europa.eu/publications/media-literacy-and-the-fight-against-disinformation [5] Danish Parliament (2021). Act on the use of artificial intelligence systems. Retrieved from https://www.retsinformation.dk/Forms/R0710.aspx?id=205797
Technology advances, such as deepfakes, are being exploited in political arenas to manipulate public opinion, deceiving voters with fake endorsements, impersonations, and manipulated images [3]. The exploitation of these deceptive practices can lead to defamation of candidates, amplified stereotypes, and societal divisions, fueling polarization and undermining social cohesion [2][3]. Meanwhile, the political landscape is addressing this issue, implementing legal frameworks, developing detection tools, promoting awareness, and advocating for transparency in political advertising [1][5]. However, the rapid evolution of technology poses challenges, with the effectiveness of detection tools questioning their longevity [1].
