Issue 13 · Nov 2–8, 2026Editions · Glossary · Archive · Method · RSS · HU / ENG

News of the AI era, curated weekly

Glossary › Society and ethics

election interference

basicwidespread

A hidden or deceptive intervention that uses unfair means – today often artificial intelligence – to try to sway voters' decisions or the outcome of an election.

In plain terms

It is a natural part of campaigns that parties seek to persuade: they put up posters, debate, and advertise. We speak of election influence when someone does this covertly, under a false name, or with lies, intending that you do not know who is speaking to you and why.

Imagine it like a football match. The cheering is loud but rule-bound. Influence is when someone spreads fake news with planted people in the stands, or sends one team's fans home with a false public announcement.

Artificial intelligence did not invent this phenomenon, but it made it cheaper and faster. What once required a room full of paid commenters can now be done by a few people using text, image, and voice generation tools, as well as autonomous posting AI agents.

The goal is not always to get you to vote for someone. Often it is enough that you stay home, or that you no longer trust anything, not even the real news.

How it works

  1. Target selection: organizers choose groups that are uncertain or easy to anger. They often use social media and purchased data for this.
  2. Content creation: AI is used to mass-produce personalized messages, fake news, forged images, or deepfake recordings mimicking a politician's voice and face.
  3. Distribution: a network of fake profiles, bots, and AI agents shares, comments on, and likes the content to make it seem that many real people think the same thing.
  4. Amplification: social media recommendation algorithms show content that triggers many reactions to even more people, moving it into real conversations and the press.
  5. Timing: the most harmful content often appears in the days before a vote, when there is no time for rebuttal.

An everyday example

Two days before the election, you receive an audio recording in a group chat in which a well-known politician allegedly discusses fraud. The voice is deceptively real, yet the recording was made with artificial intelligence, and the refutation does not reach everyone before the vote.

Who is affected

everyoneAs a voter, you are the target: you encounter misleading content in your news feed and chat groups.
lawyer, citizenThe integrity of elections is a constitutional issue, and the EU's new rules impose obligations on platforms and political advertisers. → Law
marketerThe targeting and transparency of political advertising are governed by strict EU rules, and violations carry heavy fines. → Marketing
developerAnyone building a content-generation tool or AI agent must account for the possibility that their system could be used for political deception. → Code

A common misconception

MythElection interference means hackers alter the vote-counting results.
In realityIn most cases, the target is not the votes but the picture in voters' minds: fake news, fake profiles, and misleading footage. There are few proven examples of hacking vote counting. Technical attacks more often take the form of leaking stolen emails or documents, which also falls under this category.

Limits and risks

Similar, but not the same

Related concepts

Broader

information manipulation, influence operation

Narrower

politikai deepfake, botnet

On the edge

disinformation, foreign information manipulation and interference (FIMI), deepfake, voice cloning, micro-targeting, recommendation algorithm, szintetikus tartalom, content labeling, AI-agent

A brief history

External and covert influence on elections is a longstanding phenomenon. Its online form came into focus after the 2016 US presidential election. In September 2023, two days before the Slovak parliamentary election, a false audio recording created with artificial intelligence spread. In it, Michal Šimečka, leader of Progressive Slovakia, allegedly spoke about election fraud. In January 2024, in the United States, AI-generated phone calls mimicking Joe Biden's voice urged New Hampshire voters not to vote in the primary. In December 2024, Romania's Constitutional Court invalidated the entire presidential election and ordered a repeat. The election was finally held again in May 2025. The decision was based on intelligence reports describing an organized online influence campaign, primarily on TikTok.

Where things stand

At the end of 2026, AI-generated deceptive content accompanies almost every major election. Authorities, platforms and fact-checkers monitor it routinely. Several EU rules apply to it. The Digital Services Act (DSA) expects large platforms to manage election risks. The AI Act requires deepfake content to be labelled. The regulation on the transparency of political advertising has applied since October 2025. As a result, several major platforms, including Meta and Google, have stopped political advertising in the EU. However, enforcement of the rules and the measurement of their impact are still immature.

Status: 2026 November

Also known as

election interference · election interference · election manipulation · foreign election interference · election interference · electoral interference · election meddling

Sources

This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.