Did the Chicago Sun-Times Use AI to Create a Summer Reading List Filled With Fake Books?

With the weather warming up, your thoughts may be turning to spending lazy summer afternoons at the beach with a good book. Tina, an online creator and co-host of the podcast Book Talk, Etc., opened her edition of the Chicago Sun-Times on Monday to find the newspaper's "Summer reading list for 2025." If Tina were me, someone who isn't in-tune with the latest in literature, this article might have seemed like a useful jumping-off point for some new book recommendations. But seeing as Tina creates content all about books, I imagine she is acutely aware of which authors are writing which books—which is why she was able to immediately spot a number of books on the list that simply do not exist.

Tina took a photo of the article and posted it to her Threads account, accusing the newspaper of using AI to generate its recommendations. That image of the article is now circulating on Bluesky, as well as the Chicago subreddit. I'm not a subscriber, and the article doesn't appear to be on the Sun-Times' website, so I can't verify the list myself. But the Sun-Times has since confirmed the list is real.

Of the 15 "books" on this summer reading list, only five are real books you can actually, you know, read: Bonjour Tristesse by Françoise Sagan; Beautiful Ruins by Jess Walter; Dandelion Wine by Ray Bradbury; Call Me By Your Name by André Aciman; and Atonement by Ian McEwan. Coincidentally, they are the last five books on the list, and are not new. The other 10 are totally made up, including such "hey, I'd read that" entries as The Last Algorithm, a new AI thriller from The Martian author Andy Weir; and Boiling Point, a smart-sounding story of environmental ethics by critical darling Rebecca Makki. Alas.

Given these fake books have real authors attached to them, fans of those authors may believe their favorite writer has a new, intriguing novel out. Even if you have no idea who any of the named authors are, you might use this piece to head to your library or book store to get a jump on your summer reading list—and if you're pulling from the first 10 recommendations, you're going to be looking for a long time.

What happened here?

According to a Bluesky post from the Sun-Times, the article was not editorial content, and was not approved or created by the newsroom. The post does not say one way or another whether the content is AI-generated, however 404 Media spoke with the author, who admitted to using AI for this article as well as others: "I do use AI for background at times but always check out the material first. This time, I did not and I can't believe I missed it because it's so obvious. No excuses."

Even before we had this confirmation, it did seem like the newspaper used generative AI to write this piece. That's not just because the writing is stilted. AI often hallucinates, or, in other words, sometimes makes things up. It's not totally clear why the models do this—it could be an issue with the training data, or the conclusions the models draw from that training—but the problem is only getting worse even as AI models ostensibly improve.

This isn't something you can avoid with better prompts, either: If you use generative AI, it's going to hallucinate sometimes, which means you need to check the outputs for inaccuracies (or straight-up lies). I follow that someone who doesn't understand this technology would see what a program like ChatGPT can do and want to use it to generate articles like this, but as many writers and artists have been arguing for years now, you can't replace a human worker with an AI chatbot and expect the same quality work. Sure, ChatGPT will happily generate you a list of 15 book recommendations in under a minute—but it's possible some (if not most) of those recommendations will be garbage.

I don't believe in using generative AI to publish stories like this. But if a newspaper is going to outsource the writing to a bot, it needs a human fact-checker (or perhaps, I don't know, an editor) to review the generation and make sure everything is correct. Though at that point, I'd suggest just just paying a human writer to offer the book recommendations themself. I guarantee you there are plenty of out-of-work or underemployed journalists who would jump at the chance. While it seems the Sun-Times has such a human writer behind the AI, that work needs to happen with each generation. If not, you get articles like this.

Out of curiosity, I asked ChatGPT for the synopsis of The Last Algorithm by Andy Weir. The bot searched the web for an answer, and, to its credit, accurately reported that the book doesn't actually exist. It made some assumptions, saying the Sun-Times definitely used AI to generate the article (though I suppose a very lazy intern who was hoping to be fired could have made up the books as well), likely because the social media posts it was pulling from suggested as much. But I also found its final thought to be particularly on-point (and accidentally self-aware):

"This incident underscores the importance of verifying information, especially when AI-generated content is involved," ChatGPT wrote.