Author Steven Rosenbaum Finds AI-Fabricated Quotes in His Book Author Steven Rosenbaum’s new book about misinformation has become a case study in the very problem it set out to examine, after he acknowledged that multiple quotes in The Future of Truth were invented by AI tools he used while writing.

Early use of AI and the discovery of “synthetic quotes”

Rosenbaum relied on AI chatbots as part of his drafting process. Only after publication did he discover that at least six quotations in the book were not real, but “synthetic quotes” generated by the systems he consulted. Another account of the episode notes that “AI put ‘synthetic quotes’ in his book,” underscoring that the inaccuracies came from AI-generated text rather than deliberate fabrication by the author.

Initially, Rosenbaum said he took “full responsibility” for the errors, according to coverage of the controversy, treating them as his own failure to fact-check the AI outputs thoroughly before they went to print.

From owning the error to blaming the bots

As scrutiny grew, including reporting in The Atlantic and Ars Technica, Rosenbaum’s framing shifted. He began explicitly blaming the AI tools, telling one outlet that the chatbots had “fucked up the book,” according to The Verge’s summary of his remarks.

That Verge piece characterizes Rosenbaum as sounding “like he’s trapped in a toxic relationship with AI,” capturing the tension between his reliance on the tools and their unreliability.

Determined to keep using AI despite betrayal

Despite the fallout, Rosenbaum says he intends to continue integrating AI into his writing. In an interview with Ars Technica described by The Verge, he called AI “a delightful writing companion… strangely creative and crafty and unusual in all these ways… and then it betrays you in ways that are just really quite horrible.” Ars’s own headline underlines the same ambivalence: “AI put ‘synthetic quotes’ in his book. But this author wants to keep using it.”

The episode highlights an emerging fault line in publishing: whether AI is a dangerous shortcut or an acceptable tool, provided authors maintain rigorous verification. For now, Rosenbaum appears intent on both warning about AI’s capacity to fabricate—and continuing to write with it.

Continue reading https://foxvector.com