Ahead of the upcoming midterms, a coordinated network of websites designed to resemble local news outlets is actively promoting Democratic candidates in key U.S. House races. An investigation by NPR revealed that when voters ask OpenAI's ChatGPT for information on competitive districts, the AI chatbot frequently cites these sites as authoritative sources. In one instance regarding Michigan's 4th Congressional District, ChatGPT identified Democrat Sean McCann as having the "most detailed published affordability plan," citing a website called the Kalamazoo Beacon as its source.
The Kalamazoo Beacon and nine other similar sites are part of a larger operation identified by digital security firm VisionHeight. This network consists of 37 websites that promote various candidates, predominantly Democrats, for federal, state, and local offices. While these sites label their content as "news articles" in their source code and describe themselves as reporting on local communities, they function as thinly veiled political advertisements. For example, the Kalamazoo Beacon has published only 16 stories, all of which focus exclusively on McCann's policy positions and legislative record.
Multiple indicators suggest a single operator controls the entire network. Registry data shows that all 37 websites were established on September 15, with the majority registered within an eight-minute window. Hundreds of articles appeared in batches during September and October, leading VisionHeight to speculate that AI may have been used to generate the content. The sites are optimized for AI consumption; they explicitly allow major AI crawlers to scrape their data and include files named "llms.txt" to list articles in plain text, a feature uncommon among standard news organizations.
The impact of this network varies significantly depending on the AI platform consulted. In tests conducted by NPR using 37 different prompts, ChatGPT cited at least one article from the network in approximately half of the responses. In 17 of those instances, the chatbot specifically referenced the sites pretending to be local news outlets. The phenomenon appears more pronounced in down-ballot races where credible reporting is scarce, a situation experts describe as a "data void." Conversely, Google's products behaved differently; Gemini did not cite any links from the network, and Google Search did not surface them on its first page results.
Experts warn that this strategy represents a new frontier for deceptive political advertising. Renée DiResta, an associate research professor at Georgetown University, noted that while optimizing messages for AI is expected, pretending to be news outlets crosses into deception. She emphasized the need for transparency regarding funding and linked efforts. Google spokesperson Davis Thompson stated that the company detected a policy violation in one instance and took swift action, noting their systems are designed to block spam and abuse. OpenAI spokesperson Nate Evans responded that ChatGPT cites a range of sources to allow users to make their own judgments, though it remains unclear if specific changes were made after NPR flagged the sites.
Attempts to identify the operators of these websites have been unsuccessful. NPR could not locate or contact the site owners. Promoted candidates, including JoAnna Mendoza of Arizona and Sean McCann, as well as major Democratic campaign committees and super PACs like American Bridge 21st Century and House Majority PAC, denied involvement or did not respond to requests for comment. The network's articles often link to external sources such as congressional voting records and partisan news outlets to substantiate claims, while opposing candidates appear infrequently, usually in contexts highlighting potential political liabilities.
Sources: NPR Politics.