DooDooLamb News
We Asked the ‘Future of Truth’ Author to Explain How He Used AI. It Didn’t Go Well
Brief published May 31, 2026 · Original source published May 29, 2026
Original reporting by Kate Knibbs at wired.com.
Automated brief. Verify important details at the original source.
Author of Book About AI's Threat to Truth Gets Caught Using AI to Make Up Truth
A man who wrote an entire book warning about artificial intelligence's dangers to authentic information just got busted for using artificial intelligence to fabricate information in that very book. The irony is so thick you could use it to train a large language model.
Steve Rosenbaum's "The Future of Truth" promises to explore how AI shapes our perceptions of reality, which is already a bit like a fish writing about the dangers of water. But things got deliciously meta when eagle-eyed readers discovered that Rosenbaum had sprinkled AI-generated quotes throughout his manifesto about AI's threat to truthfulness. It's the literary equivalent of a fire safety instructor accidentally setting the building ablaze during his presentation.
The controversy erupted after WIRED published an excerpt from the book, because apparently 2024's version of peer review happens in real-time on social media. Readers quickly identified quotes that seemed suspiciously polished, coherent, and free of the usual verbal tics that make human speech recognizably human. Think of it as the uncanny valley effect, but for quotation marks. When pressed about these suspiciously articulate passages, Rosenbaum's explanations evolved faster than a startup's pivot strategy during a funding crisis.
Initially, Rosenbaum attempted what we might charitably call the "enhanced transcription" defense. He claimed he had merely cleaned up interview recordings using AI, transforming the natural "ums," "uhs," and meandering tangents of human speech into crisp, thesis-ready soundbites. This is a bit like saying you only used Photoshop to remove a few blemishes from your dating profile pictures, when you've actually replaced your entire face with Ryan Gosling's. The practice sits in that special gray zone that modern technology loves to create, somewhere between editorial enhancement and outright fabrication.
But the real comedy gold emerged when it became clear that some quotes weren't just polished versions of real conversations, they were entirely AI-generated fabrications. Rosenbaum had apparently asked his artificial research assistant to imagine what various experts might say about AI's impact on truth, then presented these synthetic musings as actual human thoughts. It's like writing a cookbook by asking ChatGPT what food might taste like, then serving the resulting recipes at a dinner party. The guests might not immediately notice something's wrong, but eventually someone's going to ask why the "authentic Italian pasta" tastes like cardboard seasoned with regret.
The deeper problem here isn't just academic dishonesty, though that's certainly part of the charm. Rosenbaum has stumbled into creating a perfect demonstration of his book's central thesis while simultaneously undermining any authority he might have had to make that argument. He's become a living, breathing example of how AI can blur the lines between authentic and artificial information, except he did it to himself, in a book about exactly this problem. It's performance art, assuming the performance is "How to Destroy Your Own Credibility in Real Time."
What makes this particularly delicious is the timing. We're living through what historians will probably call the "everyone's figuring out AI ethics on the fly" era, where new moral frameworks are being stress-tested daily by people who probably should have thought things through first. Publishers, authors, and readers are all trying to establish norms around AI use in creative and scholarly work, and Rosenbaum just provided a masterclass in how not to do it. He's like the person who tests whether the ice is thick enough by jumping on it repeatedly until it cracks.
The publishing industry, meanwhile, finds itself in the awkward position of having to retroactively establish standards for something that didn't exist when most of their contracts were written. How do you fact-check a quote when the "source" might be a hallucinating language model? What's the difference between using AI to transcribe an interview versus using AI to imagine what an interview might have sounded like? These are the kinds of questions that keep publishers awake at night, right after "Will anyone still read books when they can just ask ChatGPT for a summary?"
For working developers and content creators, this mess serves as a cautionary tale about the importance of being transparently honest about AI assistance. The technology is useful, powerful, and here to stay, but pretending it's something it's not will eventually bite you in ways that hurt more than just your pride. The correct approach isn't to avoid AI tools, it's to use them responsibly and disclose that usage in ways that don't make your readers feel like they've been sold a counterfeit Rolex.
The real tragedy is that Rosenbaum probably had something valuable to say about AI's impact on truth and reality. But now his book will be remembered primarily as an unintentional case study in exactly the problems he was trying to warn people about. He's achieved a kind of literary ouroboros, where the medium became the message by completely devouring itself. Future authors take note: when writing about the dangers of AI-generated misinformation, maybe don't use AI to generate misinformation. It's a simple rule, but apparently not simple enough.