Tracking uncovers $80 million spent on AI political ads, most without disclosure

In the 2026 U.S. midterm elections, AI-generated political ads are appearing more frequently on television screens and social media platforms. Researchers from the Wesleyan Media Project have tracked approximately $80 million in spending on nearly 170 unique AI-powered ads. Their findings indicate that Republican candidates and interest groups are more likely to employ AI in their advertisements compared to Democrats.

The research team has discovered several deepfakes featuring prominent politicians, such as Donald Trump, Nancy Pelosi, Barack Obama, Kamala Harris, and Alexandria Ocasio-Cortez. Deepfakes have been used to create hyperrealistic videos depicting politicians engaging in actions they did not perform, potentially misleading voters. Republican sponsors, including candidates and interest groups, are responsible for 80% of the ads and 83% of the spending.

AI-generated ads often feature crowds or constituents, and some advertisers enhance visuals rather than create entirely new content. This makes it difficult for viewers to distinguish between AI-enhanced and non-AI-generated ads. While some advertisements explicitly disclose the use of AI, many do not, particularly in states without laws requiring such disclosures.

Polling suggests that 78% of registered voters favor banning AI content that makes deceptive claims about candidates. However, the patchwork of state legislation governing AI use in political advertising makes tracking and enforcing disclosure requirements challenging. Researchers have found that, regardless of state laws, the actual use of disclaimers in AI ads remains inconsistent.