A recent analysis of congressional campaign websites has revealed widespread use of artificial intelligence (AI) tools in the creation of candidate biographies and position statements. Investigators examined the online content of 695 candidates running for the U.S. House and Senate and found that at least 291 websites contained text likely generated by AI.
The study employed two AI-detection tools, Pangram and GPTZero, which analyze linguistic patterns such as character frequency and word order to identify content produced by AI. While these detectors are not infallible, Pangram has demonstrated accuracy rates exceeding 99 percent in common online writing, according to a 2025 University of Chicago study.
Among the flagged websites was that of Eric Pratt, a Republican candidate for Congress in Minnesota, whose campaign briefly included an unusual internal note on its site urging a focus on local politics rather than former President Trump. Pratt’s campaign acknowledged the use of AI tools for refining text. Another example included the biography of Ken Paxton, the Republican Senate candidate in Texas, whose website contained language identified as being fully AI-generated. Paxton’s campaign did not respond to requests for comment. By contrast, the website for Paxton’s Democratic opponent, James Talarico, was determined to have human-written text exclusively.
The use of AI has provided campaigns with a swift and cost-effective method of producing polished prose. Dale Pabst, campaign manager for Patrick Halpin, a Democrat running against Republican incumbent Andrew Garbarino in New York, noted that AI assisted in assembling the campaign platform within days rather than weeks, helping to level the playing field despite a significant financial disadvantage.
Experts note that AI tools, such as OpenAI’s ChatGPT and Google’s Gemini, generate text by predicting common writing patterns, which can lead to repeated phrases and predictable styles across different websites. For instance, a common phrase flagged in multiple campaigns was structured as “X isn’t just… it’s…” appearing in posts touching on issues ranging from environmental concerns to economic topics.
While over 30 states have enacted laws requiring disclosure when AI is used in campaign advertisements—especially to combat deceptive deepfake videos—no regulations currently mandate candidates to reveal AI usage in written materials. Political operatives have increasingly incorporated AI behind the scenes, leveraging it for voter data management, drafting campaign messages, and processing feedback, with some even using chatbots to assist in campaign strategy.
The incorporation of AI in campaigns raises questions about transparency and authenticity, especially amid ongoing legislative efforts to regulate AI technology and its underlying infrastructure. Many of the campaigns identified with AI-generated text were lower-profile races with limited funding, though some were involved in high-stakes, heavily financed contests.
As AI tools become further embedded in the political process, their influence on how candidates communicate with voters may grow, prompting calls for clearer guidelines and disclosure practices in future elections.