Nieman Lab: AI fake news complaining about how AI fake news is the death of real news
Nieman Lab’s Joshua Benton documented a specific case in July 2026 that illustrates how far AI content fabrication has developed. A spam operation called “The Editorial” had been producing fictional news articles about American newspapers closing—complete with fabricated quotes, invented publication names, specific dates, and bylines written to mimic real journalists. The detail that gave the piece its recursive quality: the fake articles were themselves about how AI fake news was destroying real journalism.
The case
The Editorial was generating content in a subject area—local newspaper closures—with genuine emotional resonance for journalism readers. Stories about local papers shutting down are real, common, and widely shared. The fake versions followed the same genre conventions: named publications, attributed sources, specific closure dates, and quotes from staff. The articles circulated on social media before being traced back to the spam network.
The recursive structure matters beyond the anecdote. AI-generated content about AI-generated misinformation is not just an ironic footnote. It signals that the disinformation ecosystem has become self-referential: the production of fake news is now consuming coverage of fake news as raw material, because that category of content is credible to and widely read by the audiences most likely to spread it.
What made the content hard to detect
Benton’s analysis focuses on what made The Editorial’s content initially credible: realistic article structure, specific fabricated details that looked like reporting, and writing that avoided the most recognizable AI phrasing patterns. Without active fact-checking on the named publications and quoted individuals, the content was difficult to identify on first encounter.
Practical implications for writers and editors
The piece has two practical implications for editorial teams. First, AI content verification can no longer rely primarily on detecting AI-like writing patterns. The patterns are becoming indistinguishable quickly enough that verification needs to return to primary sources: does this newspaper actually exist, did this person actually say this, can this date be confirmed. Second, the choice of subject matter matters in disinformation. Content that targets emotionally resonant topics within a trusted audience’s area of professional concern will spread further than generic AI content, which means editorial teams should be more alert in exactly the subject areas they care most about.