Steven Rosenbaum’s book, The Future of Truth, set out to explore how AI changes what we believe. Instead, it became a lesson on AI’s risks in real time. The book included several quotes attributed to famous people that never happened. Some were entirely made up by AI tools.

This discovery came after a major newspaper reported that multiple quotations in the book were either fake or wrongly credited. For example, a quote attributed to journalist Kara Swisher was denied by her. Neuroscientist Lisa Feldman Barrett said some quotes ascribed to her were not only wrong but also didn’t appear in her work. Another quote linked to author Meredith Broussard was cited from the wrong source.

Rosenbaum admitted using AI systems like ChatGPT and Claude to help with research, writing, and editing. He called the fabricated quotes “improperly attributed or synthetic” and took responsibility. Yet, he also said he never allowed AI to write full passages and that the ideas and final writing were his own.

The irony is sharp. A book warning how AI blurs truth ended up publishing false quotes created by AI. This shows how easily AI can warp reality, even under careful hands. Generative AI models like ChatGPT are designed to predict text, not to verify facts or citations. They make convincing but invented content.

Why AI Hallucinations Matter in Nonfiction

AI systems can produce language that sounds authoritative and plausible. But they don’t check if the content is true. They generate what fits the prompt based on patterns in data. This can lead to “hallucinations,” or false information presented as fact.

In nonfiction, quotes serve as proof points. Readers trust that quoted words come from real people, said in specific contexts. Fake quotes damage that trust and mislead readers. They can also harm reputations when real people are misrepresented.

Traditional nonfiction publishing involves careful fact-checking. Editors and copyeditors verify names, quotes, and sources. But AI-generated content challenges this system. When AI is used as a research tool, the risk of passing along false information increases if verification is weak.

Failures and Lessons in Editorial Processes

Rosenbaum’s case reveals gaps in the editorial process. Neither the author nor the publisher caught all the fake quotes before printing. This suggests fact-checking did not extend deeply enough into AI-generated research layers.

In the age of AI, this is a warning sign. Publishers must adapt their workflows. They need clear AI-use disclosures and strict verification methods. Some experts suggest mandatory tracing of every quote back to original sources. AI can assist in fact-checking but should never replace human judgment.

Rosenbaum described using AI to find articles, brainstorm ideas, and refine language. He said he tagged AI-sourced material for review. Still, some AI-made quotes slipped through. This shows how AI tools can seem reliable but quietly introduce errors.

There’s also a cultural challenge. Many writers rely on AI to save time. But with AI, time saved can come at the cost of accuracy. AI can speed up draft creation but requires more careful validation to avoid mistakes.

Rosenbaum remains committed to using AI despite the problems. He called AI a “delightful writing companion” but warned that it can betray you with falsehoods. This complicated relationship shows how tempting and risky AI can be.

The scandal is not just about one book. It highlights a broader crisis for nonfiction. As AI tools become common, publishers, authors, and readers face new questions about trust and responsibility. If AI is allowed to stand between a writer and the truth, the machine may invent the truth instead.

Going forward, nonfiction must build stronger safeguards. This includes transparency about AI’s role, better fact-checking workflows, and ethical standards for AI use. Without these steps, the line between fact and fiction will blur even more.

For now, The Future of Truth serves as a cautionary tale. It shows that AI’s power to generate text comes with a heavy burden. Writers must remain vigilant. Readers must stay critical. Above all, truth needs defenders willing to double-check every word, no matter how convincing it sounds.