Steven Rosenbaum's book The Future of Truth explores how AI distorts reality, yet a New York Times investigation found the very problem he documented: synthetic quotes that never existed. Rosenbaum used tools like ChatGPT and Claude to gather information, tagging AI-generated notes with warnings. However, six problematic citations slipped through, including quotes that Kara Swisher and Lisa Feldman Barrett say they never made.

Rosenbaum describes his relationship with AI as simultaneously intoxicating and dangerous, acknowledging he learned a hard lesson about trusting outputs. He plans to continue using AI tools despite the errors, calling the technology 'magical' for connecting ideas and providing new thinking pathways. The author argues AI assisted with research—surfacing ideas and summarizing themes—but maintained human control over narrative structure and conclusions.

This incident reveals how traditional fact-checking fails with AI-assisted research. When fact-checkers assumed quoted material was verifiable written text, they missed that AI can fabricate authoritative-sounding quotes that don't exist. Publishers may need new verification workflows requiring source tracing for quotations and better provenance tracking.

The irony isn't lost on Rosenbaum—his book warning about AI's reality-bending effects became a case study in those same failures. He frames this as instructive, demonstrating how persuasive these systems have become even to critics.