The impact of disinformation generated by AI on democracy case studies: the US presidential elections in 2016 & 2024 Open Access

The Internet has vastly increased global access to information but also enabled the spread of false and insignificant content. Digital platforms prioritize engagement over quality, leading to a rise in disinformation. This poses unprecedented threats to democracy and human rights by confusing citizens, undermining trust in institutions, disrupting elections and fostering skepticism about important issues like climate change (Colomina et al., 2021, pp. 2–3). Western democracies face challenges from cyberattacks and information operations aimed at manipulating public opinion. These issues are exacerbated by societal tensions and the rise of technologies like Artificial Intelligence (AI), which have changed how information and disinformation are created and disseminated (Kertysova, 2018, pp. 55–56). A decade ago, social media was hailed as a tool for empowering citizens and enhancing democracy, exemplified by its role in the Arab Spring. However, it turned out to be a threat, particularly with the rise of AI and big data. Events like the American election in 2016 and the Brexit referendum in 2016 raised ethical concerns about data privacy and disinformation. The emergence of deepfake technology further complicates the landscape, allowing the creation of deceptive videos that can undermine public trust and the integrity of information (Kaplan, 2020, p. 150).

AI technologies, particularly deepfake technology, pose a significant threat to electoral security. Deepfakes can create highly realistic videos of politicians, making them indistinguishable from real footage. These videos can be generated easily, even on smartphones, using just a few images. These fake videos can have a profound impact on elections and can change voters' choices. Deepfakes can distort a politician's stance or damage their credibility, especially during extended early voting periods. For instance, a deepfake depicting a politician with an anti-drug platform using drugs could severely affect voter perceptions and be difficult to refute quickly. Experts warn that these tactics are particularly effective during times of public uncertainty, which can lead to increased polarization and unrest (Ray, 2021, pp. 985–987).

AI technologies pose threats to the integrity of elections in two ways: First, by manipulating political discourse through social bots, which amplify opinions and spread disinformation on social media. Their effects were noted significantly after the 2016 US elections. Second, through deepfakes, which create convincing audio and video forgeries that can damage reputations and distort reality (Thiel, 2022). Another threat is Online Political Microtargeting, a strategy that uses extensive data to tailor political messages to specific demographics in order to mislead them. These developments put the democratic process at risk as they distort communication and erode trust in electoral integrity (Kreps and Kriner, 2023). The historical roots of disinformation in elections date back to the Cold War, when it was used by governments and political parties in both democracies and autocracies. In general, disinformation is used as a tool to divert attention from accurate information, making it harder for citizens to form informed beliefs about electoral conduct. It is also used by elites such as presidents, who are particularly persuasive, to significantly shape public perceptions of elections. As a result, the spread of disinformation can erode trust in democratic processes and institutions (Mauk and Gromping, 2023, pp. 972–973).

This paper examines the impact of disinformation generated by AI on the democratic process – particularly in the context of elections – using the US presidential elections in 2016 and 2024 as case studies. It also attempts to answer the following question: Why could disinformation generated by AI pose threats to democracy?.

The study will depend on the comparative case study method, relying on two case studies: (I) the 2016 US presidential election and (II) the 2024 US presidential election. This comparison would help in identifying the similarities and differences concerning the influence of disinformation generated by AI on democracy, and whether it poses a serious threat to democracy or plays a minor role in affecting the democratic process and changing voters' opinions.

Additionally, the study will use a qualitative approach, which will be based on analyzing key arguments concerning the impact of disinformation and AI on democracy, using the US elections as an indicator of democracy and the 2016 and 2024 US campaigns as platforms for examining the influence of AI-supported disinformation on electoral outcomes. The study would also present a conceptual framework that would analyze the concepts of democracy, disinformation and AI, and explore the relationship between them in a way that could contribute to a theoretical understanding of the role of disinformation enhanced by AI in affecting democracy.

Before exploring the complexities of democracy, it is important to recognize the fact that the concept does not have one comprehensive definition. Instead, it is multidimensional and varies in meaning depending on its context and interpretation. Some philosophers describe it as an “essentially contested concept”, often rooted in competing theories. Robert Dahl based his definition of democracy on three key normative principles: First, The Principle of Intrinsic Equality – all individuals are fundamentally equal, and no one has the right to impose authority over another. Second, The Presumption of Personal Autonomy – individuals should be considered the best judges of their own interests unless proven otherwise. Third, The Strong Principle of Equality – every adult member of a group is qualified to participate in collective decisions affecting their interests. He also formulated five criteria for a political process to be considered democratic: effective participation (ensuring equal opportunities for expressing preferences), voting equality at the decisive stage, enlightened understanding of relevant issues, equal opportunity to control the democratic agenda and inclusiveness of all adult members (excluding those outside the decision-making community) (Berisha, 2017, pp. 20–21).

Intelligence is fundamental to human civilization, distinguishing us from other species and enabling us to shape our environment and advance through technology. The term “artificial intelligence” (AI) combines “artificial”, indicating human-made, with “intelligence” – a concept that, while intuitively understood, is complex and not fully defined by scientists or philosophers. “Tegmark” defines intelligence as the ability to achieve complex goals, acknowledging its broad nature. Intelligence encompasses various traits, such as learning and problem-solving. Some machines, like simple calculators, can outperform humans in specific tasks, while others, like image or speech recognition systems, may still struggle with tasks that even young children can perform easily. AI can be categorized into narrow (or weak) intelligence, which focuses on specific tasks, and broad (or general) intelligence, which encompasses a wider range of goals. “Super-intelligence” refers to AI that surpasses human intelligence. Thus, AI is a multifaceted concept, with narrow AI already integrated into daily tools like Google and Facebook, while strong AI (or artificial general intelligence) remains a theoretical aspiration that inspires both visionary ambition and cautious concern (Timberg, 2017).

There is no universally accepted legal definition for the concept of “disinformation”. Ongoing discussions among practitioners and academics focus on whether a singular, comprehensive definition is necessary or practical. For instance, the European Union defines disinformation as “verifiably false or misleading information that is created, presented and disseminated for economic gain or to deceive the public intentionally and may cause public harm” (European Commission, 2018). Motivations behind disinformation can include economic interests, reputation management or political agendas. Anyone with a social media account – governments, companies or individuals – can create and spread disinformation. UNESCO distinguishes between those who fabricate disinformation (instigators) and those who disseminate it (agents), such as influencers or organizations. The most significant threats to democratic integrity arise from organized campaigns that use multiple platforms, often linked to state actors or political groups (Bontcheva and Posetti, 2020). Disinformation involves the intentional creation and spread of false information to sway public opinion. While disinformation campaigns undermining democracy have a long history, modern information technologies and social media enable the rapid and large-scale dissemination of information. In many fragile democracies, the absence of strong institutions – such as independent media and responsive political parties – hinders efforts to combat disinformation. Academics have noted that fake news and social bots contribute to the radicalization of anti-democratic sentiments, highlighting the threats posed by digital propaganda and electoral espionage in democratic societies (Martin-Rozumiłowicz and Kužel, 2019).

After examining each concept individually, it is important to synthesize them to analyze how they interact and influence one another. According to the World Economic Forum' s Global Risks 2024 and 2025 reports, disinformation is considered a significant threat to humanity, particularly in the near future. The growing dominance of digital platforms has fueled a sharp rise in misleading information, making reality increasingly difficult to discern and undermining transparency. As the reports highlight, such developments may facilitate the rise of authoritarian regimes and weaken existing democracies (Rego and Weber, 2025). Political campaigns have long involved disinformation and false claims about opponents. However, emerging technologies such as AI now pose a new and significant threat to voters and democratic processes. AI-generated deepfakes, convincing but deceptive visuals and synthetic voices have become widespread tools with the potential to influence election outcomes. Although some states are working on laws to regulate AI in political contexts, technological advancement continues to outpace both legislation and public understanding. This technological shift presents challenges to democracy, especially within a context already marked by eroding public trust, political polarization, disinformation and declining voter engagement. AI may exacerbate these challenges by enabling manipulation, surveillance and the concentration of power among technological and political elites (Lee et al., 2024). When examining the relationship between democracy and AI, it is important to recognize that this relationship can be both positive and negative. On one hand, AI can be used to mislead the public and erode democratic norms by undermining the integrity of institutions and restricting citizens' access to balanced public discourse and the freedom to form independent opinions. AI can also threaten personal privacy by enabling the unauthorized collection and analysis of individuals' data. On the other hand, democratic governance may benefit from AI through enhanced transparency, improved efficiency and more responsive public service delivery. As generative AI continues to evolve, the implementation of forward-looking policies will be necessary to ensure that core democratic principles are upheld (European Parliamentary Technology Assessment, 2024).

After examining the relationship between democracy, disinformation and AI, it becomes evident that AI functions as a double-edged sword for democratic systems. Based on this dual nature, the study seeks to examine the relationship between democracy, AI and disinformation within the context of the U.S. presidential elections of 2016 and 2024. The study will explore whether AI contributed significantly to the dissemination of false information via social media platforms – thereby undermining democratic norms – or, conversely, whether it played a constructive role in enhancing public awareness and protecting electoral integrity, ultimately reinforcing democratic principles.

The 2016 U.S. presidential election underscored the increasing influence of digital transformation on democratic processes. Algorithms, automation and AI played pivotal roles in amplifying disinformation campaigns and enabling cyber activities. These technologies not only shaped public opinion and influenced voting behavior but also demonstrated how digital tools can be exploited to manipulate electoral outcomes. As AI continues to be integrated into everyday technologies, it is expected to empower malicious actors with greater capabilities to infiltrate governmental and corporate networks, steal sensitive data, compromise individual privacy and undermine democratic institutions – often with minimal traceability (Howard et al., 2018).

The spread of “fake news” during recent social and political events – particularly the 2016 U.S. presidential election – has emerged as a major concern. Fake news, defined as fabricated or grossly distorted information, was widely disseminated through social media platforms, prompting the World Economic Forum to identify digital disinformation as a significant technological and geopolitical risk in its 2013 report. Numerous studies have investigated the role of disinformation across platforms such as Facebook, Twitter, YouTube and Wikipedia. Research on Twitter, for instance, found that false news tends to spread more rapidly and widely than factual news, primarily due to its novelty. A growing trend of political polarization in news consumption has also been observed, with individuals increasingly gravitating toward content that reinforces their existing political beliefs. Automated accounts, or bots, have been identified as major drivers in the spread of disinformation. During the 2016 election, bots were instrumental in promoting false narratives by targeting influential users with replies and mentions, while fact-checking content was rarely circulated within the core networks. One notable study analyzed 171 million tweets from 11 million users, comparing the dissemination of fake news via digital platforms with content from traditional, fact-based news outlets. The findings revealed that 10% of tweets linking to news articles directed users to fake news websites, while 15% pointed to highly biased sources. Significantly, among tweets posted through non-official Twitter clients, the proportion of users sharing links to fake news websites was more than four times higher than those linking to traditional news outlets. This suggests that bots played a disproportionately large role in the diffusion of disinformation compared to traditional media (Bovet and Makse, 2019). Some journalists and scholars have even speculated that fake news may have contributed to Donald Trump's electoral victory. These developments raise serious concerns regarding the quality of information accessible to voters and its potential implications for democratic processes. However, despite the growing attention to the phenomenon, there remains limited empirical understanding of how individuals consume fake news from untrustworthy sources and how such exposure may influence political attitudes and behavior (Guess et al., 2021).

In the three months leading up to the 2016 U.S. presidential election, misleading and false information on Facebook had a significant impact on users. The top 20 fake news stories on the platform received more engagement – measured by comments, shares and likes – than the top 20 stories from reputable news outlets such as The Huffington Post, The New York Times and The Washington Post combined. Furthermore, studies found that three-quarters of individuals who encountered such fake news believed it to be true. This demonstrates the manipulative power of textual disinformation. The potential for manipulation becomes even more severe with the advent of deepfake technology, which uses AI to generate or alter audio and video content. Deepfakes can convincingly depict public figures, such as Donald Trump or the Pope, appearing to make statements or endorsements they never actually expressed – ranging from controversial political positions to fabricated events. This highlights the escalating threat of AI-driven disinformation in shaping public opinion and undermining democratic discourse (Kaplan, 2020, p. 155).

There are some insights that show how fake news affected Donald Trump's 2016 presidential victory like: “Fake News Sharing”, in the three months leading up to the election, fake news stories supporting Trump were shared 30 million times on Facebook, while those supporting Clinton were shared 8 million times. “Belief in Fake News”, it was found that over half of the American adults, who remembered seeing fake news stories, believed them to be true. “Confirmation Bias”, people were more likely to believe fake news stories that favored their preferred candidate, especially when they were part of ideologically homogenous social media networks. “Social Media's Role”, social media is considered an important source of information, but it was not the dominant one, as only 14% of Americans reported that social media was their “most important source” of election news. “Youth and Fake News Perception”, in 2015 a Stanford study found that over 30% of high school students trusted a fake Facebook account posting pro-Trump comments more because of its graphic design. Additionally, most students struggled to recognize signs of credibility, such as the blue checkmark on Twitter or Facebook, indicating legitimate accounts. More than 80% could not distinguish between sponsored content (native advertising) and real news stories. These findings suggest that while social media and fake news were influential, their full impact on the election outcome remains difficult to quantify. However, they underscore the challenges of navigating the complex media landscape, especially for younger and less media-literate audiences (Georgacopoulos and Mores, 2020).

One of the most prominent incidents illustrating the impact of AI and cyber operations on the 2016 U.S. presidential election was the widespread belief in the involvement of Russian hackers during the campaign. The Democratic Party publicly stated that it had been targeted by cyberattacks. A confidential CIA report later confirmed that Russian operatives were behind these attacks, with the intent of aiding Donald Trump in securing both the Republican nomination and ultimately the presidency. Trump's election was widely perceived as aligning with Russian geopolitical interests, as he advocated for a more inward-looking and less interventionist U.S. foreign policy. Throughout his campaign, Trump pledged to reduce U.S. commitments to NATO and avoid involvement in issues concerning Ukraine and Crimea. His repeated admiration for Russian President, Vladimir Putin, further reinforced perceptions of a pro-Russian stance. As a result, Trump's victory was viewed as a potential challenge to longstanding pillars of U.S. foreign policy, particularly the global promotion of democracy and American values (Parliamentary Assembly of the Mediterranean, 2024, pp. 1–2). The disinformation campaign orchestrated by Russia during the 2016 election highlighted the power of cyberwarfare and digital disinformation to influence political outcomes – capabilities that could not have been achieved as effectively through traditional military means such as invasion or bombing (Padda, 2020, p. 3). Russian operatives deployed advanced technologies in a coordinated strategy to manipulate public opinion and disrupt democratic processes.

According to a 2018 report by the U.S. Senate Intelligence Committee, various digital tools – including targeted advertisements, fabricated news stories, self-generated content and social media platforms – were strategically employed to manipulate and deceive millions of American users. The primary objective was to polarize the U.S. public along societal, ideological and racial lines, provoke real-world unrest and covertly bolster Russia's preferred candidate in the 2016 election (Kreps and Kriner, 2023). The scope of this influence campaign was extensive. Russian-purchased Facebook advertisements reportedly reached an estimated 10 million users. Furthermore, Russian operatives created 470 fake Facebook accounts, with just six of those accounts generating content that was shared more than 340 million times. Research conducted by Columbia University's Tow Center for Digital Journalism and New Knowledge, along with findings from the Mueller Report, revealed that the Russian-financed Internet Research Agency (IRA) spent approximately $100,000 on over 3,500 Facebook ads between June 2015 and May 2017. These ads predominantly conveyed anti-Clinton and pro-Trump messages. Initially, Facebook denied that fake news had significantly impacted the election and claimed no knowledge of Russian-funded advertisements on its platform. However, the company later admitted that Russian operatives had published approximately 80,000 pieces of content over a two-year period, which collectively reached an estimated 126 million Americans. Despite Facebook's assertions that it had implemented significant reforms to combat disinformation and enhance platform transparency, serious concerns remained. In August 2019, a coalition of philanthropic organizations studying the platform's impact on democracy threatened to end collaboration with Facebook, citing its failure to provide researchers with promised access to critical data. This incident underscored persistent doubts about the company's accountability and the effectiveness of its efforts to prevent political manipulation (Hynes, 2021, pp. 9–12).

While the Russian disinformation campaign during the 2016 U.S. presidential election operated on an unprecedented scale, it was not without flaws that may have limited its overall effectiveness. One notable weakness was the presence of subtle linguistic errors in many social media posts – particularly in the form of awkward phrasing or grammatical mistakes, such as misplaced or omitted articles – that were atypical for native English speakers. These anomalies often served as cues that the content was inauthentic or foreign-generated. Nevertheless, emerging evidence suggests that AI-generated propaganda can be just as persuasive as that created by human actors. When paired with advanced microtargeting techniques, such as those leveraging users' behavioral data and preferences, AI-driven disinformation campaigns could dramatically increase in precision and influence. These technological developments imply that future operations may be significantly more sophisticated, scalable and difficult to detect than those employed in 2016 (Vincent, 2023).

The Trump and pro-Brexit campaigns shared several key connections, particularly among influential figures and organizations that played a major role in shaping both movements. One significant link is the involvement of individuals like “Richard Mercer”, an early AI pioneer and billionaire, who supported both campaigns, along with other prominent figures such as “Peter Thiel”, a wealthy entrepreneur and Trump supporter and “Steve Bannon”, a key figure in the right-wing Breitbart News Network who later became a strategist in Trump's administration. A central component of both campaigns was the use of data-driven political strategy, largely facilitated by Cambridge Analytica. This company was pivotal in both the Brexit campaign and the Trump election campaign, employing sophisticated psychological profiling and microtargeting techniques to sway voters' opinions. Many of the people mentioned earlier, including Mercer, Thiel and Bannon, were either investors or partners in Cambridge Analytica, further tying these influential figures to the manipulation of public opinion through data-driven methods. The use of AI and digital strategies in both the Trump and Brexit campaigns, reflects a coordinated effort to leverage technology for political advantage (Anderson and Horvath, 2017). Cambridge Analytica used a similar approach in both the Brexit referendum and the 2016 presidential election in the US, focusing on data-driven methods to target voters. However, a particularly concerning strategy employed more rigorously by the Trump campaign was voter suppression. Reports suggest that the campaign targeted three groups likely to support Hillary Clinton – white liberals, young women and African Americans – by creating ads designed to discourage them from voting. While it is difficult to measure how effective these ads were, the strategy itself raises serious democratic concerns, particularly if these efforts succeeded in reducing voter turnout among these key demographics (Green and Issenberg, 2016).

The year 2024 has been described as “the year of elections and global campaigns”, with pivotal elections taking place not only in the United States but also in the European Union and major countries such as Brazil, India, Croatia and Mexico. Collectively, these nations represent over 41% of the global population and 42% of global GDP, underscoring the global significance of electoral outcomes in 2024. However, these elections are unfolding amid unprecedented technological challenges – chief among them, the rapid advancement of generative AI. Generative AI, which can produce highly realistic text, images, audio and video from simple prompts, has progressed remarkably in recent years. Since the viral launch of ChatGPT in November 2022, there has been growing concern about the potential for such technologies to intensify the spread of disinformation. Commentators have even warned of a looming “disinformation nightmare” in the lead-up to the 2024 elections (Wirtschafter, 2024). Public awareness of the risks posed by AI-generated disinformation is also increasing. A global Ipsos survey conducted in spring 2023 found that over 60% of respondents believed AI-generated deepfakes could convincingly depict politicians in false and damaging ways. In addition, AI-driven fake accounts are capable of spreading disinformation at scale, while image generation tools can be used to create overly flattering or misleading portrayals of political candidates, thereby distorting public perception (Mackenzie and Scott, 2024). These concerns are compounded by ongoing threats of foreign interference. Coordinated disinformation campaigns, often powered by bots and online trolls, continue to amplify political polarization, promote hate speech and distort democratic discourse. As a result, technology is making electoral campaigns more complex, divisive and susceptible to manipulation than ever before – posing serious risks to the integrity of democratic processes worldwide (Dommett, 2023).

The proliferation of AI-generated content, particularly deepfakes, is emerging as a critical challenge in the context of the 2024 U.S. presidential election. These sophisticated synthetic media technologies are becoming increasingly difficult to detect, thereby amplifying the potential for disinformation and public manipulation. According to “Simon Horswell”, a senior fraud specialist at Onfido, there was an alarming 3,000% increase in deepfake attempts in 2023 alone – an indication of the rapidly escalating scale of the problem. This exponential growth suggests a corresponding rise in the threat posed to electoral integrity. Dr Ilia Kolochenko, CEO of ImmuniWeb, warns that generative AI can produce millions of “malicious brainwashing messages” that can be disseminated widely and instantaneously through social media platforms. These AI-generated messages not only exploit users' cognitive biases but also contribute to an environment in which distinguishing fact from fabrication becomes increasingly difficult. As a result, the growing sophistication and accessibility of generative AI tools are enhancing the effectiveness of disinformation campaigns and threatening the foundations of democratic decision-making (Lee, 2024).

The rapid advancement of AI, in the absence of robust regulatory frameworks, presents pressing moral and technical challenges – particularly in the context of governance and electoral processes. This unregulated landscape, often described as the “Wild West” of AI, is heightening public concern and posing serious risks to the integrity of democratic elections. First, the emergence of AI-generated deepfakes – including fake audio, video and images – has significantly accelerated the spread of disinformation, particularly via social media platforms. A notable example occurred in January 2024, when a robocall in New Hampshire used an AI-generated voice clone of President Joe Biden to falsely urge voters to skip the Democratic primaries. Such incidents illustrate how synthetic media can directly undermine voter trust and distort the democratic process. Second, AI has expanded the capacity for both domestic and foreign actors to interfere with electoral infrastructure. This includes targeting voter registration databases, voting machines and public opinion through microtargeted psychological operations. A precedent was set during the 2016 U.S. presidential election, when international hackers used AI-assisted techniques to breach the Democratic National Committee's network, leak confidential documents and weaken public confidence in the electoral system. Third, AI systems can perpetuate bias through flawed algorithmic design, potentially resulting in unequal treatment or disinformation directed at specific voter groups. Fourth, the mass data-processing capabilities of AI pose a serious threat to privacy, as electoral systems often handle sensitive personal information, such as voter registration details and political affiliations – making them lucrative targets for cyberattacks (AI's Impact on the 2024 Presidential Election, 2024).

In the context of the 2024 U.S. presidential election, technologists and election experts have warned that disinformation is expected to play a significantly greater role than in previous election cycles – particularly when compared to the 2016 election. In the weeks leading up to the 2024 vote, the nature and scale of disinformation became increasingly evident. According to “Tim Harper”, senior policy analyst for democracy and elections at the Center for Democracy and Technology, false claims targeting key political figures emerged, including fabricated allegations of sexual misconduct against vice-presidential nominee “Tim Walz” and a deepfake video depicting election officials destroying ballots. Both instances were linked to coordinated Russian disinformation efforts (Gross, 2024a).

Cody Buntain, an assistant professor at the University of Maryland's College of Information, emphasized that much of AI's influence on elections remains largely invisible to the public. A major function of AI lies in the algorithmic curation of content across social media platforms such as TikTok, X (formerly Twitter) and Facebook. These AI-driven algorithms personalize the user experience by prioritizing specific types of content, thereby shaping the information users encounter and engage with during the election period. A study by the Pew Research Center revealed that a significant number of Americans now rely on social media as their primary source of news, which underscores the critical role AI technologies play in shaping political discourse and public opinion. The convergence of this dependency on social media with the increasing sophistication of AI-powered disinformation mechanisms heightens the risk of distorted voter perceptions and potentially manipulated electoral outcomes (Gross, 2024b).

The platform X (formerly Twitter), now owned by Elon Musk, has adopted a markedly different approach to content moderation, heightening the risk of election-related disinformation during the 2024 U.S. presidential election. Musk – who has himself shared election-related falsehoods – oversaw the dismantling of critical trust and safety mechanisms, including the layoff of moderation staff and the removal of tools previously used to identify and flag misinformation. In their place, the platform now relies more heavily on “Community Notes,” a crowdsourced system that allows users to add contextual annotations to posts. While intended to encourage transparency, this shift has raised concerns about the reliability and speed of fact-checking, especially in politically charged contexts. Moreover, Musk reinstated numerous previously banned accounts, including those suspended for spreading election disinformation, further enabling the amplification of false narratives. According to cybersecurity expert “Chester Wisniewski”, X may have “discredited itself enough” that its content is no longer widely trusted by the public; however, its reach and influence remain significant (Leingang, 2024). Adding to these concerns, some users on X who regularly disseminate election disinformation, AI-generated images and conspiracy theories have claimed that they receive financial compensation – sometimes amounting to thousands of dollars – from the platform. These individuals operate semi-coordinated networks that support varying political agendas, including both pro-Trump and pro-Kamala Harris narratives, as well as independent messaging. Importantly, they assert no formal affiliation with political campaigns, highlighting the increasingly blurred line between decentralized social media influence and organized political efforts (Spring, 2024). These developments illustrate how structural and financial incentives on social media platforms may contribute to the erosion of democratic integrity in the digital age.

The proliferation of sophisticated generative AI tools has introduced new dimensions to political disinformation, particularly in the context of the 2024 U.S. presidential election. These technologies can now produce cloned human voices and hyper-realistic images, videos and audio within seconds and at minimal cost. When deployed alongside powerful social media algorithms, such content can be rapidly disseminated and precisely targeted to manipulate specific audiences – taking electoral interference to unprecedented extremes. Former President Donald Trump utilized AI-generated content during his 2024 campaign, including a manipulated video of CNN host Anderson Cooper. Created using an AI voice-cloning tool, the video distorted Cooper's reaction to a CNN town hall event featuring Trump, misleading viewers about the broadcaster's stance. Other viral examples of AI-generated political disinformation included a fabricated video of President Joe Biden appearing to make anti-transgender remarks, as well as doctored images depicting Trump resisting arrest – images whose creator later admitted they were fake (Klepper and Swenson, 2023).

One of the most persistent disinformation themes in 2024 centered on immigration and border security. Donald Trump repeatedly claimed that migrants were “overrunning” the U.S. southern border, depleting public resources and contributing to rising crime rates. These exaggerated or misleading claims fueled highly polarized narratives, intensifying political divisions and shaping public opinion through fear-based messaging. In this deeply divided climate, surveys revealed that some Americans knowingly shared false information online, reflecting a troubling normalization of disinformation in political discourse. If such trends continue, they may significantly undermine trust in democratic institutions and further erode the stability of American governance (West, 2024).

A 2024 survey conducted by the Pew Research Center revealed widespread public concern in the United States regarding the role of AI in the presidential election. According to the findings, 39% of Americans believed that AI would be used primarily for harmful purposes during the campaign, while only 5% expected it to be used positively. Another 27% anticipated a mix of both positive and negative impacts. A clear majority – 57% of U.S. adults – expressed high levels of concern that AI would be utilized to create and spread fake or misleading information about candidates and political campaigns. Notably, this concern transcended partisan divisions, with 56% of Republicans and 58% of Democrats sharing similar anxieties. Public expectations of major technology platforms were also high: 77% of Americans believed that companies such as Facebook, X (formerly Twitter), TikTok and Google had a responsibility to prevent the misuse of their platforms in ways that could influence the election. However, public confidence in these companies' ability to meet that responsibility has declined sharply. Only 20% of respondents expressed confidence in tech companies' ability to prevent misuse, down from 33% in 2018. The data indicate a significant bipartisan consensus on the potential for AI to negatively influence democratic processes, as well as growing skepticism about the tech sector's commitment or capacity to safeguard electoral integrity (Pew Research Center, 2024).

Senator Mark Warner, Chair of the Senate Intelligence Committee, warned that the 2024 U.S. presidential elections are likely to be more susceptible to foreign disinformation than previous electoral cycles. He identified several contributing factors to this heightened vulnerability, including the evolution of more sophisticated disinformation tactics by foreign actors such as Russia and China, the increasing role of domestic political groups in amplifying false narratives and the rapid development of advanced AI tools capable of generating convincing fake images, audio and video content. Warner also pointed to a troubling trend among major technology companies, many of which have scaled back their efforts to combat disinformation, further weakening defenses against manipulation (Klepper, 2024). In parallel, recent intelligence assessments and cybersecurity reports have indicated a sharp increase in cyber operations targeting the United States, particularly by Russia, China and Iran. While cyberattacks are not new, recent patterns suggest a more coordinated and strategic approach aimed at undermining U.S. governance and economic stability ahead of the 2024 election. Russia, traditionally viewed as the primary cyber threat, continues to probe vulnerabilities in American infrastructure. Meanwhile, China appears to be taking a more observational role – possibly preparing to contribute to a broader disinformation or cyber campaign – while Iran has also ramped up its cyber activities. These developments point to a potential alignment among adversarial states seeking to destabilize the U.S. political environment at a critical democratic juncture (Mosser and Cox, 2024).

Foreign interference in U.S. elections – most notably Russia's influence operations during the 2016 presidential race – continues to pose a significant and evolving threat. By 2024, the integration of AI into these interference efforts has further escalated their potential impact. In July 2024, the U.S. Department of Justice announced the seizure of two web domains and the dismantling of nearly 1,000 social media accounts operated by a Russian “bot farm”. This campaign bore striking similarities to Russia's 2020disinformation operations but now featured enhanced AI capabilities that allow for greater scale, speed and targeting precision. AI-generated content can mimic authentic discourse, making detection more difficult and manipulation more effective. In addition to Russia, Chinese actors have also been implicated in AI-driven disinformation campaigns targeting the U.S. political system. One notable example involved a falsified transcript of a speech by President Joe Biden, which falsely suggested he made inappropriate sexual remarks. The altered transcript was circulated on social media platforms, aiming to generate public outrage and sow distrust. These incidents underscore how generative AI can be weaponized by foreign adversaries to distort political narratives, influence voter perceptions and undermine democratic institutions (Trish, 2024).

Besides Russia, the Chinese-language social media played a great role in spreading disinformation, fueled by the rise of AI. Most of this disinformation has focused on promoting Trump and Republican policies, while attacking Biden and Democratic policies. The study will mention three of the most famous Chinese false narratives spread around the 2024 US presidential election. The first narrative: “Trump as the Savior of the U.S.”, a key disinformation narrative painted Donald Trump as the only person capable of saving the U.S. from its current crises. Supporters of this narrative portray Trump as unjustly targeted and emphasized the idea that his leadership is essential for restoring American values, inflating his popularity and public support in the process. The second narrative: “Biden, Harris, and the ‘Extreme Left’ bringing the U.S. to Communism”, claimed that the “extreme left” policies of Joe Biden, Kamala Harris and the Democratic Party are destroying the United States and pushing the country toward communism. In particular, disinformation about Harris focused on sexist and misogynistic attacks. This narrative also attacked Biden administration for causing economic problems, such as inflation and rising crime. The third narrative “The “Great Replacement” Conspiracy: Migrants Voting for Democrats”, it claimed that Biden administration intentionally opened the southern border to allow migrants to vote in the 2024 election. The disinformation claimed that Biden administration has intentionally relaxed border security, provided aid to migrants and used executive actions to grant them voting rights (Place, 2024, PP. 9–13). These disinformation campaigns were particularly potent because they resonated with the existed political fears and stoke racial and cultural anxieties. The spread of these narratives on Chinese-language social media platforms aimed to distort voters' perceptions of candidates, undermine confidence in the election process and create threaten the whole democratic process.

An examination of the impact of disinformation generated by AI during the 2016 and 2024 U.S. presidential elections' campaigns reveals both striking similarities and notable differences in the nature and scope of foreign interference and technological manipulation.

In both elections, disinformation – primarily driven by foreign actors – played a pivotal role in influencing voter behavior and undermining trust in democratic institutions. In each case, the campaigns largely benefited Donald Trump, with Russia identified as a central actor in spreading false narratives to polarize American society. The strategic deployment of bots and fake accounts to amplify misleading content, was a shared tactic across both election cycles. Moreover, social media platforms outpaced traditional media in the dissemination of disinformation, serving as the primary channels through which manipulated content reached the public. The underlying objective in both elections remained consistent: to exploit societal, ideological and racial divisions, erode trust in democratic systems and manipulate public opinion. These efforts contributed to widespread anxiety about election security, with bipartisan concern over the integrity of the electoral process. Following both elections, policymakers and analysts acknowledged the growing threat that AI poses to democracy, particularly in terms of information reliability and the manipulation of voter perception.

Despite these parallels, several important differences distinguish the 2024 election from its 2016 predecessor – primarily due to advancements in generative AI. By 2024, AI technology had evolved significantly, enabling the creation of hyper-realistic images, audio and video within seconds and at minimal cost. This marked a substantial leap from the relatively basic use of bots and algorithmic targeting in 2016. The widespread adoption of tools like ChatGPT, introduced in 2022, further exacerbated the potential for misinformation by making it easier to produce persuasive, human-like content at scale. Platform dynamics also shifted. In 2016, Facebook played a dominant role in the spread of disinformation. By 2024, platforms such as X (formerly Twitter) and TikTok emerged as influential vectors for disinformation, particularly in disseminating AI-generated deepfakes and manipulated content. Additionally, social media's influence on voter opinion intensified in 2024, with a significantly larger portion of the American public relying on these platforms as primary sources of election-related information. Another notable difference is the scale and coordination of cyber warfare. While Russia remained a major player in 2024, China and Iran also became prominent sources of AI-augmented disinformation, signaling a broader, more coordinated approach to destabilizing U.S. democracy. Finally, while tech companies responded to the 2016 interference with efforts to improve moderation and algorithmic transparency, many of these safeguards were relaxed by 2024, increasing the system's vulnerability to manipulation.

Despite the clear risks associated with AI-fueled disinformation, the same technology also holds potential to strengthen democracy if used ethically and strategically. AI can be employed to detect and mitigate false content in real time, enhance media literacy and ensure transparency in electoral processes. These tools could support fact-checkers, identify coordinated inauthentic behavior and alert users to manipulated media.

The following section will explore how the U.S. government has responded to the risks posed by AI in the 2024 election cycle. It will also offer policy recommendations and technical solutions to harness AI's capabilities in safeguarding democratic processes and countering the threat of disinformation.

In response to the growing threat posed by AI-generated disinformation, U.S. policymakers at both the federal and state levels have begun to implement regulatory measures aimed at safeguarding the electoral process. At the federal level, Congress introduced at least five legislative proposals focused on the use of AI in elections, with two of these bills advancing through the Senate Rules Committee as of May 2024. Additionally, independent regulatory bodies such as the Federal Communications Commission (FCC) and the Federal Election Commission (FEC) are actively exploring the extent of their authority to address AI-related challenges in the electoral domain (McIsaac, 2024, p. 8). These efforts reflect a growing awareness among lawmakers of the urgent need to contain the risks associated with generative AI technologies. At the state level, 17 states have taken legislative action to address the use of AI in elections. These laws range from outright bans on certain deceptive AI applications to requirements for disclosure when AI-generated content is used in campaign materials. Although these measures represent a step forward, experts warn that such regulatory efforts may not be sufficient to entirely neutralize AI's influence on electoral outcomes. Beyond formal legislation, election integrity also depends on proactive efforts by both election officials and voters. Election officials play a critical role in countering disinformation by serving as credible sources of information. By building trust long before election day and ensuring communication remains transparent and consistent, these officials can help protect the public from being misled by false narratives created through AI-generated content. Likewise, voters themselves must adopt a critical stance toward the media they consume, particularly in the final weeks of an election cycle when disinformation campaigns tend to intensify. Encouragingly, surveys suggest that a growing number of voters are developing skepticism toward AI-generated information – particularly from tools such as ChatGPT – and are more cautious about accepting unsolicited election-related content at face value. This informed skepticism, coupled with coordinated efforts from election authorities, can form a crucial line of defense against the spread of AI-driven deception in democratic processes (Ozdemir, 2024).

The US government becomes more aware of the potential impact of AI on the election process, which demands careful attention and safeguards to protect the integrity of democratic systems. Here are some strategies that were proposed to ensure the legitimacy of U.S. presidential elections: First, Legislation and Regulation like “The For People Act”, which seeks to expand voter access, limit the influence of money in politics, strengthen ethical standards and protect election systems from interference. It includes measures to modernize voter registration, require paper ballots and audits and establish a national commission to safeguard democratic institutions. Second, Cybersecurity Measures, in response to potential cyber threats, the Department of Homeland Security has labeled election systems as critical infrastructure. The Cybersecurity and Infrastructure Security Agency (CISA) provides election officials with cybersecurity resources and guidance to protect election integrity. Third, Transparency and Accountability, coalition of tech companies, civil organizations and researchers has developed guidelines to ensure responsible AI use in elections. These guidelines include disclosing sources of AI-generated content, ensuring data accuracy and allowing for independent audits. Fourth, Public Education, which requires campaigns to be launched in order to educate voters about AI's role in elections. These include quizzes to help identify AI-generated content, tools to combat disinformation and online resources providing trusted election information. Finally, International Cooperation, The U.S. and European Union have created a joint working group to enhance election security, share best practices and coordinate responses. The United Nations has also addressed AI's impact on peace and security, promoting ethical standards for AI use based on human rights principles (AI's Impact on the 2024 Presidential Election, 2024).

AI itself can be used effectively to combat disinformation during election times, as it has been used to identify fake bot accounts through techniques like bot-spotting and bot-labeling. By labeling these accounts, social media platforms help users understand which content may be coming from automated sources, allowing them to better assess its reliability. However, the accuracy of detection algorithms still requires improvement to match the effectiveness of technologies like email spam filters. Major companies such as Google, Facebook and Twitter use machine-learning algorithms to combat online disinformation by identifying and removing fake accounts, trolls and harmful content. For instance, Facebook reports that AI tools are responsible for detecting 99.5% of terrorist-related content removals, 98.5% of fake account removals, 96% of adult nudity and sexual content and 86% of graphic violence removals. These AI systems are often trained with data from human moderators. Building on this success, Facebook is trying to expand its use of AI to identify and flag false news stories, as well as detect duplicates of already debunked content, further enhancing the platform's ability to combat disinformation (Kertysova, 2018).

The evolution of AI has significantly transformed the landscape of electoral politics in the United States, intensifying both the opportunities and risks associated with digital campaigning. A comparison of the 2016 and 2024 U.S. presidential elections reveals both continuity and escalation in the use of AI-driven disinformation, underscoring the urgency of systemic safeguards to protect democratic integrity.

In both elections, foreign interference – particularly from Russia – played a central role in spreading disinformation with the intent to polarize voters and destabilize trust in democratic institutions. Social media served as the primary vehicle for this manipulation, with both cycles witnessing the strategic use of bots and fake accounts to amplify misleading narratives. Public concern was evident in both years, with bipartisan anxiety about election integrity and the harmful impact of disinformation on voter perception. Yet the 2024 election differed markedly in scale, sophistication and the nature of the technological threats involved. While 2016 relied heavily on coordinated disinformation through text-based posts and memes – largely on Facebook – the 2024 cycle was characterized by the widespread use of generative AI to create hyper-realistic deepfakes, fabricated audio clips and synthetic news stories across multiple platforms, including TikTok and X. The disinformation threat expanded beyond Russia to include China and Iran, reflecting a more complex and multi-actor environment. Moreover, public reliance on social media for political information had significantly increased by 2024, amplifying the potential reach of AI-generated content. Notably, while tech companies had introduced protective measures after 2016, many of these were scaled back by 2024, exposing new vulnerabilities in platform governance. Despite fears of catastrophic disruption, AI's impact on the 2024 election was more limited than some anticipated, due in part to heightened public skepticism and improved digital literacy. However, the growing realism of AI-generated content and its ability to outpace fact-checking efforts suggest that future elections may face even greater risks unless proactive measures are adopted.

Therefore, AI must be seen not only as a threat but also as a tool. With responsible use, it can support election integrity – through automated detection of fake content, bot identification and voter education tools. To achieve this, political actors must commit to transparency, tech companies must enhance platform accountability and civil society must foster public resilience to manipulation. The comparison between 2016 and 2024 reveals a clear trajectory: disinformation is evolving, becoming more automated, more persuasive and more deeply embedded in digital ecosystems. Safeguarding democracy in the age of AI requires moving beyond reactive responses to a sustained, cooperative effort – across governments, industries and societies – to ensure that truth can still shape political decision-making in the digital age.