Author Steven Rosenbaum says an external investigation identified several quotes in his new book, The Future of Truth, that lack verifiable sources or appear to be fabricated, and he now acknowledges “a handful of improperly attributed or synthetic quotes.” He told a news outlet he and his publisher are conducting a full citation audit and preparing corrections for future editions — an outcome that matters because the book’s central theme is how AI reshapes truth.
The investigation singled out two named examples: a line attributed to tech reporter Kara Swisher that she says she “never said,” and material tied to neuroscientist Lisa Feldman Barrett that she says “don’t appear in [my] book, and they are also wrong.” Rosenbaum says these flagged passages stem from research notes and that the errors slipped through the editorial process.
Rosenbaum says he used OpenAI’s ChatGPT and Anthropic’s Claude extensively to gather information, tagging AI-mined passages in his notes as “this came from AI,” and passing those notes to a fact-checker plus two copy editors provided by the publisher. He describes the tools as useful “to surface ideas, locate articles, summarize themes, identify people or papers I might want to look into,” while insisting the reporting, narratives and arguments in the book remain his own.
Among the book’s 285 outside citations, the investigation identified six problematic entries, including three so-called synthetic quotes with no apparent source. Rosenbaum concedes the editorial pipeline missed those errors despite double — checking and warned more examples could surface as the work undergoes further review: “I think we did that [double — checking] incredibly effectively, but not a hundred percent.
The episode highlights a practical limit of traditional fact-checking when AI is introduced into research workflows. Historically, fact-checkers could verify a quoted written source by checking the original text; Rosenbaum and others note that AI-generated summaries or invented quotes break that assumption, creating a need for additional verification steps and provenance tracking when tools like ChatGPT or Claude are used.
Rosenbaum says the experience has changed his approach — he is “going to be much more suspicious” and “reticent to trust” AI outputs — yet he rejects an AI-free process, calling the tools “magical” for knitting ideas together. He and his editors are conducting a systematic citation audit and preparing corrections for future editions, underscoring a takeaway for authors, publishers and toolmakers about the need for clearer provenance and extra skepticism in AI-assisted research.
Sources
Replies (0)
No replies in this topic yet.
