Democrats sent fake newspapers to voters' homes for years to deceive them before Election Day.
They just upgraded the scheme and pointed it straight at ChatGPT.
When voters asked the chatbot who cared about their community, it had only one name ready.
How Democrats Built Fake Local News Websites to Feed ChatGPT Misinformation
Someone registered 37 websites on September 15 – 36 of them in an 8-minute burst – designed to look like legitimate local news outlets covering competitive House races across the country.
Ten of those sites posed as local journalism and each one covered a single Democrat candidate in a competitive House race.
The Kalamazoo Beacon in Michigan described itself in its website code as covering southwest Michigan and had published exactly 16 stories, every one of them about Democratic House candidate Sean McCann.
When a researcher asked ChatGPT which candidates in Michigan's 4th Congressional District cared about affordability, the chatbot cited the Kalamazoo Beacon and declared McCann had the "Most detailed published affordability plan."
That answer came from a site that has never covered anyone else.
Each of the 37 sites was built from scratch to fool chatbots, not readers.
Legitimate news organizations block automated bots because they pull articles without paying for them.
These sites did the opposite – they named 17 different bots explicitly and welcomed every one in, including those operated by OpenAI, Google, Meta, and Apple.
Each candidate got a profile running thousands of words alongside a voter guide of more than 50 questions written to match the exact way people search on ChatGPT.
Across the network, researchers found one phrase repeated more than 360 times – and on a single page dedicated to one candidate, it appeared 34 times: "is the stronger choice."
Each profile was embedded with hidden code classifying it as a news article, which is how ChatGPT ended up steering voters toward candidates based on what were effectively campaign brochures.
Digital security firm VisionHeight uncovered the operation and found the detail that blew it wide open.
Each of the 10 so-called local news sites shared an identical description buried in their own website code: "Straight local news – reports on the race and (once a content-sourcing workflow exists) community storylines, in neutral third-person news style."
They hadn't even finished building the disguise – and it worked anyway.
ChatGPT cited these sites in 19 of 37 test queries.
In Michigan, when researchers asked ChatGPT about five city council candidates covered by one of the network's sites, the chatbot cited the fake outlet every single time.
Google caught it – Gemini didn't cite any of the fake sites and Google Search didn't surface them either. ChatGPT fell for it, and OpenAI still hasn't fixed the problem.
Democrats Have Run This Dark Money Fake News Scheme for Years
The fake website network is a newer version of a con with a long history inside the Democrat Party.
Democratic operative David Brock built networks of locally branded fake news sites pushing pro-Democrat messaging during the 2023 Virginia elections – the same concept, before the reach was global.
A Soros-funded dark money operation mailed a fake newspaper to nearly one million Michigan voters before the 2024 Senate primary – one that falsely called Democrat Senate candidate Abdul El-Sayed a "veteran."
Nebraska Democrats launched a fake local newspaper before a Senate race with no public disclosure about who funded it – the exact definition of the "dark money" they claim to despise.
The 2026 version is the same fraud at a fraction of the cost. No printing bills, no postal routes – just a template, a few hundred domain registrations, and a laptop. Instead of reaching mailboxes, it goes straight into the tool millions of Americans now use to decide how to vote.
The Party That Cried Misinformation Built an AI Election Interference Machine
Democrats spent years building their entire political brand around fighting misinformation.
They pressured Big Tech to censor stories they disliked, pushed government agencies to flag conservative content for removal, and spent years accusing Republicans of poisoning the news.
The whole time, they were running one of the most sophisticated propaganda operations in American political history – one designed specifically to exploit the technology Americans trust most for neutral answers.
What makes it worse is who caught them.
NPR – the formerly publicly funded outlet that spent years carrying water for the Democrat Party – stumbled into exposing this while testing chatbots ahead of the midterms.
Nobody knows who funded the 37 sites. The DCCC denied involvement when NPR called, and the super PACs – American Bridge, Tomorrow Together, and House Majority PAC – didn't respond at all. That's exactly how they designed it.
The party that invented "dark money" as a political talking point built an untraceable propaganda machine, disguised it as local journalism, fed it to artificial intelligence, and watched ChatGPT recommend their candidates to voters who thought they were getting neutral answers.
Democrats didn't just deceive voters – they programmed the machine to do it for them.
Sources:
- David Strom, "NPR Reveals Democrat Scheme To Rig AIs Into Deceiving Voters," Hot Air/ZeroHedge, November 1, 2026.
- Washington Free Beacon Staff, "Soros-Funded Fake News Site Falsely Identified Abdul El-Sayed as 'Veteran' in Mailer Sent to Michigan Voters," Washington Free Beacon, 2024.
- Washington Free Beacon Staff, "Dark Money Dems Use Fake Newspaper To Influence Virginia Election," Washington Free Beacon, 2022.
- Flatwater Free Press Staff, "Out of the Blue, Left-Leaning Dark Money Group Launches Nebraska Newspaper Ahead of Election," Flatwater Free Press, 2026.
