We Requested Insights from the Author of ‘The Future of Truth’ on His Use of AI, but the Results Were Unfavorable.

Earlier this month, WIRED shared a passage from Steve Rosenbaum’s highly discussed new book, The Future of Truth, which examines how artificial intelligence distorts people’s perception of reality. Not long after, The New York Times revealed that the book included several fabricated or misattributed quotes. In response, Rosenbaum, who holds a master’s degree in “truth” from New York University, acknowledged the inclusion of “a handful” of “improperly attributed or synthetic” quotes. Ironically, the credibility of a book about AI’s influence on truth was now under scrutiny due to the author’s use of AI.

Following the Times article, WIRED revisited our 1,450-word excerpt. The fact-checking team had ensured its accuracy before publication, and we confirmed that the quotes and facts were valid. However, WIRED’s generative AI editorial policy forbids the publication of AI-generated content, and a reader email labeling the excerpt as “blatantly AI-written” brought up further concerns about the extent of Rosenbaum’s reliance on AI tools. In the acknowledgment section of The Future of Truth, Rosenbaum mentions using ChatGPT, Claude, NaturalReaders, ProWritingAid, and Grammarly to “refine and polish the presentation of [his] ideas.” What did this entail, exactly?

WIRED utilized several AI-detection tools, such as Pangram, GPTZero, and ZeroGPT, to analyze the excerpt. Each service indicated that it was likely AI-generated or generated with high confidence. However, since AI-detection tools have their limitations and can yield inaccurate results, WIRED’s head of research reached out to Rosenbaum directly to inquire about his use of AI in crafting the excerpt.

He responded, stating, “Like many writers today, I used AI tools during various stages of the research and editorial development process for the book, including source discovery, brainstorming, structural feedback, and language refinement.” He emphasized, “The ideas, reporting, arguments, and final authorship are mine, and the WIRED excerpt was not generated by AI and published as-is.” He advised WIRED’s editors to exercise caution when relying on AI detection tools, mentioning the possibility of false positives.

At this stage, WIRED’s senior editors tasked me with investigating the incident, as I have reported on AI in its many forms since 2024. My initial action was to run the entire book text through Pangram’s detection tool. (While all AI-detection tools have shortcomings and may show false positives, Pangram is currently the gold standard.) It indicated that the book was approximately 53 percent AI-generated, with another 9 percent categorized as likely AI-assisted.

I contacted Rosenbaum to obtain a more detailed account of his AI usage while writing the book and whether he disputed Pangram’s findings. (BenBella Books, the publisher of The Future of Truth, did not respond to requests for comment. Simon & Schuster, which distributes BenBella’s books in the U.S., also declined to comment.)

Rosenbaum refrained from commenting on the accuracy of Pangram’s findings and seemed reluctant to discuss them at all. “I don’t engage in that conversation,” he stated. “It’s akin to saying, do you beat your wife? It’s a type of accusation that lacks a proper response.”

Instead, he offered a broad explanation of his editorial approach. He indicated that he initially used AI tools as search engines to gather information for the more research-intensive sections of the book. To illustrate this, he asked ChatGPT to describe me and then read the AI’s findings aloud. The AI accurately summarized some of my previous work, including coverage on AI-generated “zombie media sites.”