Artificial Intelligence (AI) in Elections

Can AI become a tool for democratic progress, or will it become a weapon for its destruction?

An image of “democracy in the era of AI” generated by Meta AI, which includes a watermark (bottom right) denoting its origins.

There is no doubt that artificial intelligence (AI) will transform democracy and democratic institutions, for better or worse.

Many academics and commentators have made predictions about what and AI-dominated future might look like. The predictions are wide-ranging—some fear AI will undermine and destroy the very trust that democracy relies on, while others believe it will unleash a new era of civic engagement.

On this page, we summarize some of the key predictions about AI and democracy. For more detail, read our report Artificial Intelligence and Democracy: For Better and Worse?

Read the report

The Two Extremes

Experts have outlined two extreme scenarios for AI's role in democracy, ranging from a dystopia to a new golden age. The "Pandora" scenario describes AI being used to harm democracy, primarily by making it easier for governments and powerful groups to control information and manipulate citizens. On the other hand, the "Jeffersonian" scenario sees AI as a tool to improve democracy by empowering marginalized people and increasing citizen involvement.

Reference: Savaget, Paulo, Chiarini, Tulio and Evans, Steve. (2019) “Empowering Political Participation Through Artificial Intelligence.” Science and Public Policy Vol. 46(3), pp. 369–380.

Election Administration

AI offers opportunities to increase efficiency, but also comes with significant risks.

Opportunities: AI could help maintain accurate voter lists, select voting locations to reduce wait times, and even help protect voters from intimidation. AI-based optical character recognition (OCR) systems are already used to count paper ballots.
Risks: The risks are numerous, including potential voter disenfranchisement from overzealous voter roll purges, bias in predictive models, and the violation of ballot secrecy.

References: Deepak P, Simoes, Stanley and MacCarthaigh, Muiris. (2023) “AI and Core Electoral Processes: Mapping the Horizons.” AI Magazine Vol. 44(3), pp. 218-239; Juneja, Prathm. (2024) Artificial Intelligence for Electoral Management (International IDEA).

The Information Environment

AI is a "threat accelerator" that exacerbates existing election interference efforts.

Disinformation vs. Misinformation

Disinformation is deliberately false information intended to deceive, while misinformation is false information shared without the intent to deceive. AI can be used to generate both, and it makes it harder for citizens to distinguish between what is real and what is not.

Examples of mis- and dis-information (not necessarily generated with A) from X (formerly Twitter). Images sourced from the Anti-Defamation League.

Deepfakes

A major concern is the rise of deepfakes—fabricated videos, audio, or images that use AI to create a convincing but false depiction of a person.

A humorous deepfake of President Biden cursing that was shared widely on X (formerly Twitter) in 2024.

The Path Forward:

A multifaceted approach is needed to combat AI-generated misinformation. Experts suggest that generative AI companies should implement rules for watermarking content, and social media platforms should monitor and flag disinformation.

A close-up of the watermark on the AI-generated image of a "democracy sausage" to the right.

References: Chowdhury, Rumman. “AI-Fuelled Election Campaigns are Here — Where are The Rules?” Nature (9 April 2024); McKay, Cameron and Trauthig, Inga. “Then and Now: How Does AI Electoral Interference Compare in 2025?” Centre for International Governance Innovation (August 7, 2025); Noti, Adav. (2024) “How Artificial Intelligence Influences Elections and What We Can Do About It.” Campaign Legal Center (28 February, 2024); Wong, Wilson. (2025) “Trends in Political Science Research: Artificial Intelligence and Voter Disinformation.” International Political Science Abstracts Vol. 75(2), pp. 205-222.

Campaigns

AI will change how political campaigns are conducted, from political advertising to the use of campaign volunteers.

Political Advertising

AI can write, design, and produce campaign materials quickly and cheaply, potentially leveling the playing field for less-resourced campaigns. However, this also means campaigns have to compete with a lot of noise.

AI as an Assistant

AI is already being used as a "campaign assistant" to draft communications and may soon take on the role of a "campaign volunteer," engaging voters in real-time "dynamic, digital conversations" using chatbots and evolving AI technologies.

Ethical dilemmas

In the absence of laws outlawing the use of AI-generated content, every campaign will need to weigh ethics against the potential political gains.

Some campaigns will choose to create or amplify content they know to be fake, like when Florida Governor Ron DeSantis’ 2024 presidential campaign shared deepfakes portraying Donald Trump hugging Anthony Fauci (see below).
Some candidates will choose to exploit the unreliable information environment, by falsely claiming that damaging stories about them are "fake news" to avoid political fallout.

Even campaigns that decide against using AI-generated content may find them selves unwittingly creating or sharing misleading content, potentially leading to scandal or embarrassment for the candidate.

References: Adeleke, Fola. (2025) “Reinstating Trust in Elections in the Era of Artificial Intelligence and Emerging Technologies.” Data & Policy Vol. 7. Foos, Florian. (2024) “The Use of AI by Election Campaigns” LSE Public Policy Review Vol. 3(3), pp. 8-15; LaChapelle, Christina and Tucker, Catherine. (2023) “Generative AI in Political Advertising.” Brennan Center for Justice Expert Brief (November 28, 2023); Lee-Geiller, Seulki. (2024) “AI Use in Election Campaigns Alters Voter Perceptions of Electoral Integrity, Driven by Partisan Biases.” SSRN papers.

Citizen Trust and Civic Engagement

AI can both erode public trust and offer new ways for citizens to access information.

Erosion of Trust

A constant flood of AI-generated content can lead to a "reasonable cognitive prophylactic measure" where citizens simply choose to believe nothing, which is corrosive to social trust and democracy.

Increasing Access

On a more positive note, chatbots, when trained on validated, official information, have the potential to increase access to information and boost civic engagement.

A screenshot of New York City's "MyCity" chatbot, a chatbot intended to help NYC residents find information. Like all emergent technologies, chatbot are not without their limitations , and the MyCity chatbot was mired in controversy when it advised business owners to engage in behavior that was contrary to the law.

References: Csernatoni, Raluca (2024). “Can Democracy Survive the Disruptive Power of AI?” Carnegie Endowment for International Peace (18 December 2024); Kreps, Sarah and Kriner, Doug. (2023) “How AI Threatens Democracy.” Journal of Democracy Vol. 34(4), pp. 122–131.