A comprehensive analysis of artificial intelligence’s actual impact on global democracy in 2024
Thousands of New Hampshire voters received calls in January 2025 that sounded exactly like President Biden urging Democrats not to vote in the state’s primary. “We know the value of voting Democratic when our votes count. It’s important you save your vote for the November election,” the voice said1. It wasn’t Biden—it was a sophisticated AI deepfake that crystallized fears about artificial intelligence’s potential to manipulate democratic processes.
Yet as 2024’s unprecedented global election cycle concluded—with 3.7 billion eligible voters across 72 countries—the AI apocalypse many predicted never quite materialized. Instead, artificial intelligence’s influence on politics proved more subtle but perhaps more pervasive: a “death by a thousand cuts” through memes, propaganda, and gradual erosion of information integrity rather than the bombshell deepfakes experts feared2.
This complex reality reveals both AI’s immediate limitations and its long-term threat to democratic discourse. While sophisticated deception remained largely confined to isolated incidents, the technology’s democratization enabled new forms of political warfare that may prove more dangerous precisely because they operate in plain sight.
The Prediction vs. Reality: When Fears Met Facts
Leading into 2024, experts warned of an AI-driven disinformation catastrophe. “The nightmare situation was the day before, the day of election, the day after election, some bombshell image, some bombshell video or audio would just set the world on fire,” said Hany Farid, a professor at UC Berkeley who studies manipulated media2.
That nightmare scenario largely failed to materialize. “It wasn’t quite the year of AI elections like many folks foretold,” observed Zeve Sanderson from NYU’s Center for Social Media and Politics2. The Biden deepfake—created by a Democratic consultant who claimed he wanted to raise awareness about AI threats—became one of the few clear examples of malicious deepfake deployment, resulting in $6 million in FCC fines and criminal charges2.
Instead, the most visible AI usage was remarkably transparent: openly artificial content shared by politicians and supporters as propaganda tools. This shift from covert deception to overt manipulation represents a fundamental evolution in how AI influences politics—one that may prove more insidious than traditional deepfakes.
Research from Harvard’s Ash Center found that while AI was “everywhere” in 2024 elections, its impact came through “memes and content whose artificial origins weren’t disguised” rather than sophisticated deception campaigns3. This transparency paradox—where AI’s influence grows through openness rather than secrecy—challenges conventional thinking about digital manipulation.
Global Case Studies: AI as Political Theater
Indonesia: Resurrecting the Dead for Votes
Perhaps the most striking example of AI’s brazen political deployment occurred in Indonesia, where the Golkar party used artificial intelligence to resurrect Suharto, the longtime dictator who died in 20082. In a video posted to social media, an AI-generated Suharto declares: “I am Suharto, the second president of Indonesia” before endorsing Golkar’s candidates and stating they would “continue my dream of Indonesia’s progress.”
The resurrection wasn’t hidden or disguised—it was promoted as a technological achievement. Soon after the campaign, Suharto’s son-in-law, who enjoyed Golkar’s support, won the presidency. The incident demonstrates how AI can blur the line between historical authority and contemporary politics, allowing parties to literally conjure endorsements from beyond the grave.
India: WhatsApp Meme Warfare at Scale
In India’s massive democratic exercise—the world’s largest with over 900 million eligible voters—AI-generated memes proliferated across WhatsApp networks. At a bazaar in Jaipur, trader Dilip told NPR his friends forwarded him AI-generated content mocking opposition leader Rahul Gandhi, including one depicting Gandhi as “a stupid thief, imagining all the money he would steal if he won”2.
The memes represented a calculated risk reduction strategy. Sahid SK, a generative AI artist in India, explained: “I think that’s the only reason we haven’t seen many deepfakes in this election. Because everybody’s afraid of legal notices.” Memes, he noted, are “a wink, not outright misrepresentation”2.
This tactical shift reveals AI’s evolution as a political tool: rather than crossing legal lines with obvious deception, sophisticated operators use obvious artificiality to achieve similar propaganda effects while maintaining plausible deniability.
United States: Mainstream Adoption of AI Propaganda
In the United States, AI-generated political content achieved unprecedented mainstream acceptance. Elon Musk, owner of X (formerly Twitter), shared a fake advertisement featuring an AI clone of Kamala Harris’s voice describing herself as “the ultimate diversity hire,” posting it without initial disclosure that the content was artificial4.
Trump himself embraced AI-generated imagery, sharing deepfake images of Black Americans supposedly supporting his campaign and AI-generated pictures of Harris in Soviet military garb2. These weren’t subtle manipulation attempts but bold propaganda statements, normalizing artificial political content among millions of followers.
The Taylor Swift controversy exemplified this new dynamic: when AI-generated images falsely showed the pop star endorsing Trump, the incident sparked widespread media coverage but ultimately strengthened Swift’s real endorsement of Harris5. The backlash demonstrated both AI’s potential and its limitations when deployed against public figures with significant media attention.
The Regulatory Response: Europe Leads, America Lags
European Union: Comprehensive Framework
The European Union has emerged as the global leader in AI election regulation, implementing a comprehensive framework combining the AI Act with the Transparency and Targeting of Political Advertising (TTPA) Regulation6. The TTPA, which took effect in October 2025, requires political advertisers to disclose AI-generated content and provide transparency about targeting practices.
The regulation’s impact was immediate and dramatic: both Meta and Google announced they would cease political advertising in the EU rather than comply with the new transparency requirements7. Meta’s withdrawal from EU political advertising, effective October 2025, demonstrates how robust regulation can force platforms to make stark choices between compliance and market participation.
The EU’s approach combines multiple regulatory layers: the AI Act prohibits certain manipulative AI practices, the Digital Services Act requires platform accountability, and the TTPA mandates political advertising transparency8. This comprehensive framework offers a potential model for other democracies grappling with AI’s political implications.
United States: Patchwork Progress
American responses have been far more fragmented. At the federal level, the FCC has taken enforcement action against specific incidents like the Biden deepfake, but comprehensive legislation remains elusive. The agency has proposed broader regulations on AI-generated robocalls but faces resistance from free speech advocates and industry groups9.
State-level action has been more aggressive. Pennsylvania’s House unanimously passed legislation requiring disclosure of AI-generated content in political advertisements, joining a growing list of states attempting to address deepfakes in elections10. However, enforcement remains challenging due to jurisdictional complexity and the rapid pace of technological change.
The contrast with Europe is stark: while the EU implements comprehensive, binding regulations, the United States relies primarily on platform self-regulation and piecemeal state laws. This fragmentation creates opportunities for bad actors to exploit regulatory gaps and forum-shop for permissive jurisdictions.
The Technology Evolution: From Elite to Everyone
The democratization of AI technology represents perhaps the most significant long-term threat to election integrity. What once required specialized technical knowledge and significant resources can now be accomplished with consumer-grade applications and minimal training.
Creating convincing deepfakes in 2020 required sophisticated software and technical expertise. By 2024, mobile applications allowed users to generate realistic fake audio and video content in minutes11. This accessibility revolution means the barrier to entry for election manipulation has virtually disappeared.
Detection technology has struggled to keep pace. While researchers have developed increasingly sophisticated tools to identify AI-generated content, the arms race between creation and detection continues to favor creators12. Current detection methods often require technical expertise and significant time—luxuries unavailable during fast-moving election cycles.
The cost-benefit analysis for malicious actors has fundamentally shifted. Traditional disinformation campaigns required substantial investment in content creation and distribution. AI tools now enable the rapid generation of personalized content at scale, making previously uneconomical attacks viable for smaller actors with limited resources.
Beyond Deepfakes: The Information Ecosystem Under Siege
AI’s most profound impact may be its contribution to what researchers call “information ecosystem pollution”—a gradual degradation of public confidence in authentic content. The mere possibility that any image, audio, or video could be artificially generated creates what scholars term the “liar’s dividend”: when everything might be fake, distinguishing truth from falsehood becomes exponentially more difficult13.
This psychological impact extends beyond individual pieces of content to affect fundamental trust in democratic institutions. When voters cannot confidently distinguish authentic candidate statements from artificial ones, the basis for informed democratic participation erodes. The result is not necessarily more effective propaganda but a more cynical and disengaged electorate.
The phenomenon intersects dangerously with existing disinformation tactics. AI-generated content amplifies traditional “cheapfakes”—misleadingly edited but technically authentic material—by creating uncertainty about what constitutes legitimate editing versus manipulation. This convergence multiplies the impact of both artificial and traditional disinformation.
Farid observed that AI’s political influence operates through cumulative rather than acute effects: “I don’t think the images were designed to be clearly deceptive, but they were designed to push a narrative, and propaganda works. I do think that there was a general polluting of the information ecosystem where people just started to give up”2.
Policy Challenges: Balancing Freedom and Integrity
The challenge of regulating AI in political contexts reflects broader tensions between free expression and election integrity. Traditional approaches to campaign regulation—focused on spending limits and disclosure requirements—prove inadequate for addressing AI-generated content that costs virtually nothing to produce and can be difficult to attribute.
Current US federal proposals include requiring watermarks on AI-generated political content and expanding FCC authority over synthetic media in political communications14. However, these approaches face constitutional challenges based on First Amendment protections for political speech, even when that speech is artificially generated.
The international dimension adds complexity: AI-generated disinformation can be created anywhere and deployed globally, making unilateral national responses insufficient. Effective regulation requires unprecedented international coordination and cooperation between platforms, governments, and civil society organizations.
Platform policies have evolved rapidly but inconsistently. While major social media companies have implemented varying restrictions on synthetic media, enforcement remains sporadic and reactive rather than proactive. The EU’s regulatory pressure has forced more aggressive platform responses, but global standards remain fragmented15.
Detection and Defense: The Technological Arms Race
The development of AI detection technology has become a critical front in defending democratic processes. Current detection methods include analyzing compression artifacts, examining temporal inconsistencies in video, and identifying unnatural eye movements or speech patterns16. However, each detection advance spurs corresponding improvements in generation technology.
Research institutions and technology companies have invested heavily in automated detection systems, but human verification remains essential for high-stakes political content. This creates a fundamental scalability problem: automated systems produce false positives and negatives, while human review cannot match the speed and scale of AI-generated content production.
Blockchain-based authentication systems offer a potential technological solution, allowing content creators to cryptographically sign authentic material. However, adoption remains limited due to technical complexity and the cost of implementation across diverse media production workflows17.
Looking Forward: Preparing for Future Election Cycles
The 2024 experience provides crucial insights for defending future democratic processes. First, the threat from AI is likely to be cumulative rather than acute—gradual erosion of information integrity rather than dramatic last-minute bombshells. This suggests that defensive strategies should focus on building resilience over time rather than just crisis response.
Second, transparency may be more effective than prohibition. The EU’s disclosure requirements for AI-generated political content could provide a model for other democracies, allowing voters to make informed decisions about artificial content rather than attempting to ban it entirely.
Third, media literacy education becomes essential as AI tools proliferate. Voters need skills to critically evaluate all content, regardless of its origin, rather than relying solely on detection technology or platform policies. This requires sustained investment in educational programs and public awareness campaigns.
International cooperation will prove crucial as AI technology continues to evolve. The 2024 experience demonstrated how AI-generated disinformation can cross borders instantly while regulatory responses remain confined to individual jurisdictions. Effective defense requires coordinated responses that match the global nature of the threat.
Platform accountability must evolve beyond current self-regulatory approaches. The EU’s success in forcing Meta and Google to withdraw from political advertising demonstrates how regulatory pressure can compel meaningful platform behavior changes. Other democracies should consider similar approaches that create real consequences for platforms that facilitate AI-generated disinformation.
Conclusion: Democracy’s Digital Reckoning
The 2024 election cycle marked a turning point in AI’s relationship with democratic politics. While the apocalyptic scenarios many predicted failed to materialize, the subtle but pervasive influence of artificial intelligence on political discourse proved perhaps more dangerous precisely because it operated in plain sight.
The lessons are clear: AI’s threat to democracy operates not through dramatic deception but through gradual normalization of artificial political content. The challenge for democratic societies is developing robust defenses that preserve free expression while maintaining the information integrity essential for informed citizenship.
The European Union’s comprehensive regulatory framework offers one model, but effective defense requires adaptation to diverse political and legal systems worldwide. As AI technology continues to evolve and democratize, the window for implementing effective safeguards is narrowing rapidly.
The 2024 experience demonstrates that democracy can survive AI’s initial assault, but only with proactive adaptation and robust institutional responses. The choice facing democratic societies is clear: adapt to the age of artificial intelligence or risk the gradual erosion of the information integrity upon which democratic governance depends.
The stakes could not be higher. As AI tools become more sophisticated and accessible, the ability to distinguish authentic from artificial political content becomes central to the survival of democratic discourse itself. The lessons of 2024 provide a roadmap, but success requires immediate and sustained action across technological, regulatory, and educational fronts.
The double-edged sword of AI in politics has been unsheathed. How democratic societies wield it—or defend against it—will determine the future of informed self-governance in the digital age.
References
- Martínez-Beltrán, Sergio and Silva, Daniella. “Venezuelan deported to El Salvador mega-prison describes brutal abuse.” NPR, July 28, 2025. https://www.npr.org/2024/01/22/1226129926/nh-primary-biden-ai-robocall
- Bond, Shannon. “How AI deepfakes polluted elections in 2024.” NPR, December 21, 2024. https://www.npr.org/2024/12/21/nx-s1-5220301/deepfakes-memes-artificial-intelligence-elections
- Ash Center for Democratic Governance and Innovation. “The apocalypse that wasn’t: AI was everywhere in 2024’s elections, but deepfakes and misinformation were only part of the picture.” Harvard Kennedy School, 2024. https://ash.harvard.edu/articles/the-apocalypse-that-wasnt-ai-was-everywhere-in-2024s-elections-but-deepfakes-and-misinformation-were-only-part-of-the-picture/
- Associated Press. “AI parody video of Kamala Harris that Elon Musk shared is misleading, experts say.” AP News, July 29, 2024. https://apnews.com/article/parody-ad-ai-harris-musk-x-misleading-3a5df582f911a808d34f68b766aa3b8e
- Bond, Shannon. “Why Donald Trump’s embrace of AI-generated images has experts worried.” NPR, August 30, 2024. https://www.npr.org/2024/08/30/nx-s1-5087913/donald-trump-artificial-intelligence-memes-deepfakes-taylor-swift
- European Parliament Democracy Observatory. “The EU’s Artificial Intelligence Act and its Impact on Electoral Processes.” September 2024. https://epd.eu/content/uploads/2024/09/AI-and-elections.pdf
- Al Jazeera. “Meta to suspend political advertising in the EU as transparency law looms.” July 25, 2025. https://www.aljazeera.com/economy/2025/7/25/meta-to-suspend-political-advertising-in-the-eu-as-transparency-law-looms
- Council of the European Union. “EU introduces new rules on transparency and targeting of political advertising.” March 11, 2024. https://www.consilium.europa.eu/en/press/press-releases/2024/03/11/eu-introduces-new-rules-on-transparency-and-targeting-of-political-advertising/
- Federal Communications Commission. “FCC Statements on Midair Collision at Reagan Washington National Airport.” 2025. https://www.npr.org/2024/05/23/nx-s1-4977582/fcc-ai-deepfake-robocall-biden-new-hampshire-political-operative
- Lancaster Online. “Pa. House unanimously passes bill to crack down on deepfakes in political campaign ads.” June 24, 2025. https://lancasteronline.com/news/politics/pa-house-unanimously-passes-bill-to-crack-down-on-deepfakes-in-political-campaign-ads/article_de3dff75-7f40-4366-9b59-becfa48a4838.html
- Associated Press. “Creating realistic deepfakes is getting easier than ever. Fighting back may take even more AI.” MSN, July 2025. https://www.msn.com/en-us/news/politics/creating-realistic-deepfakes-is-getting-easier-than-ever-fighting-back-may-take-even-more-ai/ar-AA1JqVFO
- Brennan Center for Justice. “How to Detect and Guard Against Deceptive AI-Generated Election Information.” May 16, 2024. https://www.brennancenter.org/our-work/research-reports/how-detect-and-guard-against-deceptive-ai-generated-election-information
- Brookings Institution. “Watch out for false claims of deepfakes, and actual deepfakes, this election year.” May 30, 2024. https://www.brookings.edu/articles/watch-out-for-false-claims-of-deepfakes-and-actual-deepfakes-this-election-year/
- National Conference of State Legislatures. “Artificial Intelligence (AI) in Elections and Campaigns.” July 9, 2025. https://www.ncsl.org/elections-and-campaigns/artificial-intelligence-ai-in-elections-and-campaigns
- Recorded Future. “2024 Deepfakes and Election Disinformation Report: Key Findings.” July 2024. https://www.recordedfuture.com/research/targets-objectives-emerging-tactics-political-deepfakes
- ABC News. “States are cracking down on deepfakes ahead of the 2024 election.” March 27, 2024. https://abcnews.go.com/538/states-cracking-deepfakes-ahead-2024-election/story?id=108517821
- Moody’s Analytics. “Election deepfakes may undermine institutional credibility.” CNBC, July 10, 2024. https://www.cnbc.com/2024/07/10/election-deepfakes-undermine-institutional-credibility-moodys.html