---

AI-Generated Ads: The 30% Threshold

The 30% figure includes a broad range of AI involvement — from fully synthetic video to AI-written scripts delivered by real humans, to digitally altered candidate voices for targeting different accent or dialect audiences. Full synthetic deepfakes representing major candidates remain rare and legally risky in the 23 states with deepfake laws, but AI-generated voiceovers, synthetic backgrounds, and AI-personalized ad copy are now standard tools across campaigns at all levels.

The cost reduction is the primary driver. An AI-generated 30-second radio ad can be produced for under $200, versus $2,000-$5,000 for traditional production. Video ads using AI elements cost 60-80% less than full production equivalents. This has enabled candidate committees and outside groups to flood lower-tier markets and niche platforms with customized content that would have been cost-prohibitive in previous cycles.

Deepfake Legislation: A Patchwork of 23 States

Twenty-three states have passed some form of legislation addressing AI-generated deepfakes in elections as of April 2026, but the laws are far from uniform. California requires disclosure of AI-generated content in political ads within 90 days of an election. Texas bans deepfakes of candidates within 30 days of an election without their consent. Minnesota passed a similar restriction that faced an initial First Amendment challenge. Several states have passed laws targeting distribution platforms rather than creators, requiring platforms to label synthetic content.

The enforcement gap is significant. State campaign finance regulators typically lack the technical capacity to identify AI-generated content at scale. The Federal Election Commission has debated requiring AI disclosure in federal campaign ads but has not finalized rules. The practical result is that most AI-generated political content goes unlabeled and unchallenged, with compliance strongest among well-resourced campaigns concerned about reputational risk and weakest among outside groups and smaller campaigns.

Micro-Targeting: The Hyper-Local 2026 Revolution

Beyond AI content generation, the targeting precision available to 2026 campaigns has reached a level that would have seemed implausible a decade ago. A congressional campaign can now serve different ads to gun owners in a specific zip code, healthcare workers in a particular hospital's catchment area, or parents of elementary school children in a given school district — all within the same House majority. The combination of commercial data brokers, voter file matching, and platform behavioral data creates micro-audiences that allow campaigns to run multiple parallel message tracks that voters in different categories never see as contradictory. The privacy implications and the democratic implications of this hyper-targeting are significant and largely unresolved by current law.