NYT's Kashmir Hill Finds AI-Written Biography of Herself on Amazon

Kashmir Hill, the New York Times tech reporter best known for her work on privacy and surveillance, this week wrote about finding an entire unauthorized biography of herself for sale on Amazon, written by a machine. She uses that personal discovery as an on-ramp to a broader data story about what has happened to the book marketplace since ChatGPT arrived.

Two business school professors gathered data on 10 million books published on Amazon over the last five years. According to the reporting, the number of e-books published per month has tripled since the release of ChatGPT, to more than 300,000 at the end of last year, from around 100,000 in 2022. Amazon told Hill its internal metrics did not show that tripling and declined to share its own figures, which is itself part of the story: nobody outside the company can independently measure how much of the catalog is now machine-produced.

The most vivid character in the piece is Bill Johns, a 70-year-old retired cybersecurity consultant with 445 books for sale on Amazon, each carrying an AI-generated photo of himself in a dark suit rather than a real portrait. "It was either that or put on a suit and take selfies," he told Hill. The professors' analysis also flipped an assumption. They expected romance to be the genre most susceptible to AI intervention, and instead it was nonfiction, where reader ratings on AI-assisted titles ran lower than on human-written ones. "As economists," they told Hill, they are "less concerned with literary quality or customer satisfaction than revenue growth and market expansion", which is a striking reframing of what is otherwise being described as slop.

The honest caveat is that this is a first-person account anchored to one academic study, and the piece does not spell out how Amazon detects or removes AI biographies about living people, what legal recourse the subjects actually have, or how much money any of this earns. Eamon Duede, a Purdue philosopher of science quoted in the article, frames the trend more sympathetically, arguing that barriers to publishing simply come down when the machine does the writing.

The forward-looking part worth watching is on the platform side. Verified-author badges, provenance metadata, and retailer-side disclosure of AI authorship do not really exist at scale today, and the first marketplace to build them credibly will get to define what an "authentic" biography of a living person even means online.