AI-Powered 'Local News' Networks: How Partisan Sites Are Infiltrating Chatbot Responses Ahead of 2026 Midterms

NPR's October 30, 2026 investigation revealed that OpenAI's ChatGPT has been citing a network of 37 websites—many masquerading as independent local news outlets—to promote Democratic candidates in competitive House races. When queried about candidates' positions on issues like affordability in Michigan's 4th Congressional District, the chatbot prominently featured Democrat Sean McCann, linking to the Kalamazoo Beacon, a site whose 16 articles all focused exclusively on him and his record. Digital forensics from VisionHeight and Erebus identified the sites as sharing operators, with nearly all registered on September 15, 2026, in an eight-minute window, optimized for AI crawling through permissive robots.txt files and structured content. This tactic extends the longstanding 'pink slime' journalism strategy, where low-cost, partisan sites mimic local reporting to exploit trust in hyper-local sources. Carnegie Mellon University research documents how such networks have scaled ad spending into the millions ahead of recent elections, often filling voids left by declining traditional local media. While similar operations exist across the political spectrum—including Republican-linked outlets cited in National Review and LA Times reporting—the scale and AI integration here highlight a sophisticated evolution in information operations. OpenAI acknowledged the issue post-NPR inquiry but initial re-queries showed persistence; broader implications include eroded public trust in AI as neutral arbiters and amplified risks of stealth influence in an era where voters increasingly consult chatbots for election guidance. Connections to prior Democratic digital tactics, such as 2012 Google result manipulation, suggest a pattern of leveraging emerging tech platforms for narrative control, though operators' identities and funding remain opaque, with Democratic campaign committees denying involvement.