Understanding the problem of political deepfakes requires more than asking whether they can deceive. Some campaigns are using AI-generated content overtly or with full disclosure. Yet the cultural impact can still be profound: normalizing images of candidates saying or doing things that never happened makes the very concept of audiovisual evidence more fragile.
Axios documented this trend in the 2026 US election campaign, including the use of an AI version of a political opponent in an ad.
The Liar's Dividend
When everyone knows a video can be synthetic, even authentic footage can be dismissed as fake. Scholars call this phenomenon the liar’s dividend: falsification technology offers a new defense even to those whose actions are genuinely documented.
The damage is therefore asymmetric. A perfect deepfake does not need to convince millions of people; it is enough that the existence of deepfakes makes establishing a shared reality more costly.
Labels Are Necessary but Not Sufficient
Flagging the use of AI allows voters to interpret content correctly, but it does not eliminate the emotional power of the imagery. A disclosed satirical photomontage can still cement an association.
Politics has always relied on caricature and montage. The difference lies in the scale, speed, and realism with which a scene can now be fabricated. Digital literacy in the coming decade will have to include an uncomfortable skill: learning not to automatically treat one's own eyes as a reliable source.
