Short answer

CISA says generative AI mainly amplifies existing election risks such as false voting information, impersonation and influence operations at lower cost. Voters should confirm voting details on official election sites, treat shocking clips with caution, and know that the EU AI Act requires deployers to label deepfakes, though bad actors will not comply.

Key takeaways

  • CISA's 2024 factsheet said generative AI would likely amplify existing election risks rather than introduce wholly new ones, at lower cost and greater speed.
  • CISA lists AI-generated voice or video used to spread false information about the time, manner or place of voting as a possible tactic.
  • CISA also lists voice cloning to impersonate election office staff and fake videos of election officials as potential tactics.
  • Article 50 of the EU AI Act requires deployers to disclose AI-generated or manipulated image, audio or video that resembles real people or events, with an exception for evidently artistic or satirical work.
  • Labels depend on honest actors, so voters still need their own checks such as official sources and the original context.

Why election fakes spread

Elections create a short window in which a single false claim can change behaviour. If voters believe a polling place has moved, the voting date has changed, or a candidate said something outrageous, harm can be done before a correction arrives. Generative AI makes the raw material cheaper: realistic voices, faces and images can be produced without special skills.

CISA’s factsheet “Risk in Focus: Generative A.I. and the 2024 Election Cycle”, written for election officials, takes a measured view. It said that for the 2024 cycle generative AI would likely not introduce new risks but may amplify existing ones, because malicious actors, including foreign nation state actors and cybercriminals, can use it to work with greater speed and sophistication at much lower cost. That framing is useful for voters too. The tricks are old: false voting information, impersonation, phishing and influence operations. The tools are new.

What CISA says to watch for

The factsheet lists potential tactics rather than confirmed events at any one election. Those relevant to voters include chatbots, AI-generated voice or videos used to spread false information about the time, manner or place of voting by text, email, social media or print. It also lists AI-generated content that increases the scale and persuasiveness of foreign influence operations, convincing fake election records, and voice cloning used to impersonate election office staff. Other items cover deepfake videos used to harass, impersonate or delegitimise election officials, and fake audio or video of officials spreading incorrect information about the security of the election.

The factsheet also records examples of known tactics used outside elections, such as fake videos of real news anchors reporting invented stories in a foreign influence operation, and deepfake videos of famous people used to push scams. This is why election fakes and scams often look alike. The same cloned voice that sells a fake investment can read a fake voting notice. See our guides on AI voice clone scams and deepfake video call fraud.

What the law says about labels

Some governments are moving toward labelling. Under Article 50 of the EU AI Act, deployers of an AI system that generates or manipulates image, audio or video that resembles existing people, places or events, and could falsely appear authentic, must disclose that it is artificially generated or manipulated. For evidently artistic, satirical or fictional work the duty is limited to disclosing that such content exists, in a way that does not hamper enjoyment of the work. Deployers publishing AI-generated text to inform the public on matters of public interest must also disclose it, unless the text has had human review or editorial control. The reproduction I read lists Article 50 as applying from 2 August 2026, and the source is an unofficial reproduction of the text, so check the current official wording and dates. Our EU AI Act consumer guide explains the wider rules.

The limits are obvious. A rule helps when the publisher is identifiable and wants to follow it. A foreign influence operation or an anonymous account will not label its fakes. Your checks have to work without a label.

Checks that work

Start with the claim’s type. Anything about when, where or how to vote should be verified only on your official election authority’s website, reached by typing the address or using a bookmark. Anything about a candidate or official that is shocking, perfectly timed or confirms what you already believe deserves extra caution.

Look for the original: who first posted it, whether established outlets report it, and whether the clip is complete or cut. Our guides to spotting AI-generated images and video and AI-generated misinformation cover visual and source clues, though no visual check is reliable on its own.

CISA recommends election staff confirm requests through secondary challenges and use a rolling passphrase during active voting periods. Families can use the same idea. If a voice message from a relative or official asks for something unusual, call back on a number you already hold. Fake election calls and texts also overlap with phishing messages that try to steal personal details, so do not enter ID or payment details on any link claiming to be about voter registration. If you meet false voting information, report it to the platform and to the election authority, and protect your own voice with the steps in protect your voice from cloning.

What you can do

  1. 01Confirm voting dates, places, ID rules and results on your official election authority's website, not from a message, post or video.
  2. 02Pause before sharing a shocking clip of a candidate or official, and look for the original source and for reporting by established outlets.
  3. 03Be cautious about calls, texts or voice messages that claim to be from an election office or campaign, and contact the office by an official number.
  4. 04Reverse-image search suspicious images and check whether the same event appears in trustworthy coverage.
  5. 05Report fake voting information to the platform and to your election authority.
  6. 06Tell family members, especially older relatives, that voice and video can now be faked.

Related reading: AI voice clone scams and the data breach tracker.

Frequently asked questions

Are deepfakes the biggest election threat?

CISA's factsheet said generative AI would likely amplify existing risks, such as phishing, impersonation and influence operations, rather than create entirely new ones. Cheaper and faster fakes make old tactics easier to run at scale.

What does the EU AI Act require for election deepfakes?

Article 50 says deployers of an AI system that generates or manipulates image, audio or video resembling real people or events must disclose that it is artificially generated, with a limited duty for evidently artistic or satirical work. The unofficial text reproduction I read lists application from 2 August 2026, so check current dates.

Will labels stop fake election content?

Not on their own. Labelling rules bind those who follow them, and people running deception campaigns are unlikely to comply. Treat a label as a helpful signal and its absence as meaning nothing.

How can I tell if a voting message is real?

Ignore the message's own links and phone numbers. Go to your election authority's website by typing its address or using a saved bookmark, and check the date, place and ID rules there.

Sources

  1. Risk in Focus: Generative A.I. and the 2024 Election Cycle, CISA, 2024-01-18
  2. Article 50: Transparency Obligations for Providers and Deployers of Certain AI Systems, EU Artificial Intelligence Act (Future of Life Institute)

This article reflects public information as of 2026-11-06. AI tools and rules change quickly; we update articles when new facts are published. Spot an error? See our editorial and corrections policy.