How to spot AI deepfakes during election campaigns and what liability applies

Election campaigns are increasingly making use of AI-generated images, video and audio, including so-called deepfakes that are becoming harder to distinguish from genuine content. Since 2 August 2026, the EU Artificial Intelligence Act has required labelling of AI-generated content, while systems placed on the market before that date have a transition period ending 2 December 2026. Failure to comply with transparency requirements can result in fines of up to €15 million or 3% of a company's global annual turnover.

Evolving threats

Aneta Gūtmane, head of the Artificial Intelligence project at Riga Stradiņš University, notes that modern technology no longer just swaps faces in video — it can clone voices, generate real-time avatars, and produce fabricated transcripts of conversations. An additional risk comes from so-called "AI slop": mass-produced, often low-quality content that overwhelms audiences with volume and diverts attention from trustworthy sources.

Tips for checking content

Rather than hunting for visual glitches, the expert recommends checking the source of information — whether a statement also appears through official channels and reputable media — using reverse image search, and watching for suspicious social media profiles. For phone calls, people should not rely on a single source; official election information is available on the Central Election Commission's website, and for suspicious calls claiming to be from relatives, a pre-agreed password can help. In video and images, unnatural background details or inconsistent lighting may still be visible, though the once-popular "three finger test" is no longer reliable. For texts and quotes, it helps to compare the language style with a person's usual manner of communication and to trace the original source.

Legal liability

Since 22 May 2024, Section 90.1 of the Criminal Law has established liability for knowingly producing or distributing false, discrediting information about a political party or a candidate for the Saeima, a municipal council, or the European Parliament using deepfake technology during the pre-election campaign period or on election day. Penalties range from imprisonment of up to five years to short-term imprisonment, probation supervision, or community service. Residents are urged to report violations to the State Police by calling 112.