I subscribe to the Chicago Sun-Times, a non-profit newspaper. This week they sent me a sixty-four page special insert, branded with the Chicago Sun-Times logo, full of LLM nonsense. Yesterday I wrote the following letter to the Chicago Sun-Times. That evening, they published this followup.

Dear Mr. Buscaglia and the Chicago Sun-Times,

The May 18th “Heat Index” summer insert feels remarkably like “AI slop”: text generated by a large language model. The unusually enthusiastic use of adjectives coats the entire insert with an oily sheen, but I’m also concerned that there may be wholesale fabrications in the text.

For instance, “Summer soundtracks” cites Dr. Daniel Levitin, in an interview with Scientific American, as saying “Music activates the brain’s reward centers in ways similar to other pleasurable experiences. When paired with meaningful activities, these songs become powerful memory anchors that can transport us back to those moments for decades”. These phrases, and substrings thereof, return zero matches on Kagi or Google. Scientific American’s archives include a January 22, 2001 article with several quotes from Levitin, but nothing like the purported quotes.

The “Summer food trends” article cites an interview in Bon Appetit, claiming Padma Lakshmi said, “What distinguishes this summer’s global food exploration is how these international flavors are being adapted to local, seasonal ingredients.” Lakshmi is published in an interview with BA, but I can’t find any use of the uncanny marketer-speak “this summer’s global food exploration”.

The same article also cites the National Ice Cream Retailers Association as forecasting “unexpected savory notes” and “hyperlocal ingredient sourcing” as the “fastest-growing segments in premium frozen desserts”. I find it hard to believe these are segments at all—nor do these phrases appear anywhere on the NICRA web site.

The “Summer reading list for 2025” recommends books like “The Rainmakers”, by Percival Everett—a real author, but as far as I can tell, not a real text. Immediately thereafter it offers “Salt and Honey”, by “Delia Owens”—again, a real writer, not a real book. I started reaching out to some of the experts cited in the insert to ask whether their quotes were real, then realized the Verge beat me to it.

These examples are just from a cursory skim; the insert fails my LLM “sniff test” on essentially every page. How did this happen? How do you print and distribute a full-color, sixty-four page special edition without anyone reading it first? Many of the articles have no byline, but those that do are purportedly by Marco Buscaglia. Was he responsible? Or is he too, like so many of the people cited in this nonsense publication, a victim of confabulatory applied statistics? This would be a great project for the Sun-Times newsroom, assuming OpenAI hasn’t eaten them yet.

Doubtfully,

—Kyle Kingsbury