Author's Book on AI and Truth Used AI to Write It. He Won't Say How Much.
Steve Rosenbaum's The Future of Truth examines how artificial intelligence warps people's perception of reality. The book itself became a case study in that problem after the New York Times reported it contained over half a dozen fabricated or misattributed quotes.
Rosenbaum admitted to including "a handful" of "improperly attributed or synthetic" quotes. He has a master's degree in "truth" from New York University.
The irony deepened when WIRED investigated how extensively Rosenbaum had used AI tools to write the book. In its acknowledgments, he lists ChatGPT, Claude, NaturalReaders, ProWritingAid, and Grammarly as having helped "refine and polish the presentation of [his] ideas."
The Detection Problem
WIRED ran the book through three AI-detection services: Pangram, GPTZero, and ZeroGPT. Pangram flagged the entire book as 53 percent AI-generated, with an additional 9 percent likely AI-assisted.
When asked directly whether he used AI to write or edit passages, Rosenbaum gave an indirect answer. He said he used AI as a search engine to surface information, then asked what he would do with the results. "Would I take it and paste it into Google Docs and then write around it and edit it up a little bit?" he said. "Probably."
When pressed on whether he had actually done that in the book, he said: "I don't remember. You're looking for a smoking gun, and there isn't one."
Defending the Practice
Rosenbaum argued that writers now live "in fear" over AI use, which he said is "not healthy for democracy." He cited a report from PR software firm MuckRack showing 82 percent of journalists use AI daily.
He called AI "the best writing partner" he has ever had and said he would rather stop writing than stop using AI. When asked if the way he used AI made the book's accuracy questionable, he pointed back to the 82 percent statistic. "If 82 percent of journalists are using AI every day, then what you're saying is you now have anxiety about the accuracy and reliability of essentially everything that is in the current media ecosystem," he said.
That statistic, however, conflates different uses. MuckRack's survey counted AI transcription and ChatGPT searches as "AI usage." Only a quarter of the journalists polled had used AI for writing assistance-a different category than secretly incorporating AI-generated sentences into drafts.
Industry Shift Underway
Opinions on AI in media are already shifting toward acceptance. Fortune actively encourages one reporter to cowrite stories with chatbots. Business Insider permits writers to use ChatGPT on drafts. Independent journalist Alex Heath gives his reporting notes to an AI agent to generate first drafts.
Resistance remains. Hachette canceled a novel's U.S. release after Pangram indicated it was largely AI-generated. The New York Times severed ties with a freelancer for using AI. A literary magazine faced backlash after publishing what critics believed were AI-generated short stories.
WIRED retracted the Rosenbaum excerpt on Friday afternoon. The publication is revising its editorial guidelines on AI, but one principle will stay the same: published work cannot be written with AI.
Rosenbaum's situation illustrates a larger tension. His prescience on digital video and user-generated content suggests his current approach could become standard practice. A decade from now, newsrooms requiring human writers to do all the writing may seem as outdated as print delivery. For now, the lines around acceptable AI use remain contested and constantly redrawn.
For writers navigating AI tools, the stakes are clear: transparency about process matters, and detection tools have real limitations. Understanding how to work effectively with AI is increasingly necessary-but so is being honest about what you're doing.
Your membership also unlocks:
