Fakes and blanks

The eve of a book launch is a day where you don't want to encounter any unwanted surprises. That's what happened to Paul Tremblay this week: his latest novel Horror Movie hits stores, but if you went to Amazon and searched for it, you might have found an additional listing for it. This book had the right title, cover art, release date, and description, but the title was a little wonky, it was in the wrong format, and it says it was authored by a James Keith.

It was a fake, and Tremblay posted a PSA to his followers: "Please don’t buy the counterfeit paperback, likely AI generated."

The product doesn't appear to have actually been generated by a generative AI platform like ChatGPT: Tremblay noted in an email that it was likely a blank book with his cover on it. While the listing for the fake book has since vanished, it's an example of a much larger issue that's cropped up in recent years: the ease to which scam products can wind up on platforms such as Amazon, and to which they can cause confusion amongst book buyers.

Generative AI platforms such as ChatGPT use use large language models (scraped from text on the web and from books) to respond to prompts in a realistic way, allowing people to ask it to generate all sorts of things. Since ChatGPT launched in November 2022, people have used it to generate all sorts of things, such as school papers, legal briefs, articles and lists on tech sites, short stories and plenty of books. In August 2023, Amazon removed five books sold through the site that were falsely attributed to Jane Friedman, and a month later, the company announced that it would limit authors to three uploads per day to prevent a wave of faked books, and now requires authors to disclose that they've used AI for their products.

In January 2024, Wired published a report that outlined the issues that generative AI systems are creating for authors, pointing to a wave of faked products where people have turned to these types of platforms to sell generated summaries of books. In March, the Author's Guild posted a blistering condemnation of the practice, noting that "we have seen hundreds of examples of how bad actors are using generative AI to produce “books” that deceive customers and drive sales away from legitimate books," and that while Amazon has been responsive to take-down requests "the fact that they are able to get through Amazon’s content filters in the first place suggests that detecting AI-aided scams is presenting a challenge."

Scams and fake products listed on Amazon aren't a new thing: this Redditor found they'd been duped back in 2021 when they purchased a copy of David Gemmell's Legend, and enterprising sellers have been able to take advantage of the ease to which someone can upload a product to Amazon to sell copies of unpublished Star Wars novels or commission an audiobook of Dune. The Authors Guild notes that where scammers once relied on networks of poorly-paid writers to generate their fake products, they can now turn to generative AI to accomplish much the same thing.

These products are designed to fool customers by appearing to be a legitimate product by scammers taking advantage of long-standing norms within the publishing industry. Increasingly, authors and publishers are discovering that scammers are able to snag a book's cover, description, and details (things that are released well in advance of the book's publication) and can either package a file full of AI-generated content with the cover slapped on it, or something as simple as a print-on-demand notebook with the stolen cover image.

That appears to be the case with Tremblay's book. He told me in an email that writer Jason Rekulak alerted him to the fake book, and that " my editor/publisher alerted their Amazon reps, and at the same time, I shared the image online and a bunch of folks reported the notebook thing as fraudulent." Tremblay's not the only person to be subjected to this: in February, tech writer Kara Swisher released Burn Book, only to have scammers produce a flood of similar books seemingly generated by AI. And earlier this month, Luis Elizondo, who's set to publish Imminent: Inside the Pentagon's Hunt for UFOs in August, discovered that some of the readers who thought that they'd preordered the book had really purchased a fake copy – a blank notebook with his book's cover image on it.

"It's maddening," Tremblay explained to me, "and one would think Amazon might want to, I don't know, figure out a way to vet that kind of stuff, prevent it from happening. But that's probably not a cost effective act, so I doubt anything will be done to prevent these notebooks or AI ripoffs of other books from flooding the site."

Amazon does appear to be responsive to takedown requests: the fake version of Horror Movie no longer appears on the site. But it's a removal that only happens once it's spotted: if a scammer uploads and is able to sell a handful of these faked products, it likely isn't until someone receives a fake product before the scam is identified and removed, and the customer has already spent (and potentially lost) their money.

There are likely some workarounds that Amazon and publishers could put into place: a verification system (think blue checkmarks) for books being sold on the platform that would signal to buyers that it's a legitimate thing. Publishers should hold Amazon's feet to the fire. Regulators should find ways to hold Amazon and scammers accountable for these scams. And Consumers play a role here as well: paying more attention to what they're buying to make sure that the copy of Horror Movie they purchased isn't Horror Movie by Paul Tremblay Notebook.

But these types of scams work precisely because consumers and the platforms they shop on don't pay close attention, and scammers have all the tools they need to make a product look legitimate to a consumer who might be blissfully unaware that these types of scams even exist.

Ultimately, this is a problem that Amazon needs to recognize and own – for its own good. If customers can't trust that the products on a platform are legitimate, they'll move onto other places to buy their books, or could stop altogether and spend their money elsewhere, which means publishers and authors will suffer the consequences. Amazon should spend more time and energy proactively combating these sorts of scams, either by developing new systems to flag and review suspicious titles, or by implementing better reviews to catch these issues before they reach buyers on the site. Hopefully, they're working on it: the existence of platforms like ChatGPT and others means that this is a problem that isn't going away.