Author Jane Friedman spotted more books on Amazon this week that falsely claimed to be written by her than ones she actually wrote. At least five books under her name were taken down from Amazon yesterday (Aug. 8) after Friedman wrote a blog post on Aug. 7 detailing her experience finding books under her name being sold on Amazon and listed on Goodreads, the Amazon-owned social media and book-logging platform for readers.
Friedman suspects that the books are AI-generated. They don’t lift her books’ copy — instead, they copy her writing style.
Friedman’s incidence illustrates how, with the prevalence of generative AI tools like ChatGPT, book scams are likely to proliferate. That comes as new generative AI tools appear seemingly every day, making it easier to generate content in the style of a certain author.
Book scamming in the age of generative AI
How did Friedman find out about the fake books under her name? Friedman said a reader wrote to her saying she was searching for Friedman’s work on Amazon and ended up buying two books. A few pages in, the reader thought, “Oh that’s interesting. Jane’s trying a new publishing experiment of some kind.” But as she continued reading the book, she questioned whether Friedman had written any of it at all, and reached out to the author. When Friedman first looked into it, she thought the problem lay only with two books, but then she headed over to her Goodreads author page and saw there were more.
Goodreads quickly took down the fraudulent books, said Friedman, after she reached out to the site. With Amazon.com, it was less straightforward. To get in touch with Amazon, the retail giant provides a form to file infringement reports that leads to prompt takedowns of fraudulent copies. But there was no separate form for her case. “When I filed the report, I didn’t think they’re going to do anything because, like, on what grounds?” she said. “It’s not a copyright violation. And I said that up front when I filed the record.”
Here’s the message from Amazon asking for trademark registration numbers.
Amazon’s one-line response suggested it was not sent by a human.
It wasn’t until after she published a post about her findings that Friedman was contacted by someone from Amazon who assured her that they would investigate the matter. Following the contact, Friedman said her books were taken down without any explanation. “That’s exactly what I would expect from Amazon,” she said. “They only act when it reaches this level of attention and publicity. So they have no established process for dealing with these situations.”
Friedman is not concerned about AI-generated books — when it comes to the quality, she said it’s just “bloviating…It’s like a student padding their way.” She said her concern with fake books is that this is potentially reputation-damaging. “People might end up buying an illegitimate book instead of my own,” she said. “But for me, it’s not a money issue. Writing is how I make my living. And people recognize my name and they trust it.”
No Amazon infrastructure for AI-generated books in place
Book-selling scams have proliferated in the age of Amazon, eBay, and Google. But, how does one address book scams in the age of generative AI, when ways of duplicating an author’s work have become sophisticated and harder to curb?
“We have clear content guidelines governing which books can be listed for sale and promptly investigate any book when a concern is raised,” Ashley Vanicek, Amazon spokesperson, said in a statement. “We welcome author feedback and work directly with authors to address any issues they raise and where we have made an error, we correct it. We invest heavily to provide a trustworthy shopping experience and protect customers and authors from misuse of our service.”
But the retail giant does not provide guidance on how writers can protect themselves from future cases, which can get tricky. How does one know the “AI-generated” book is using that author’s name and not the author? Or what if it’s simply another Jane Friedman?
Friedman now is looking into trademarking her name, which would give her grounds to go to Amazon and say, “Take this down, here’s my trademark registration number.” Friedman suggests that writers who experience similar cases take their case to the Author’s Guild (of which she is also a member), which advocates for free expression and copyright protection and has successfully contacted Amazon to escalate complaints in the past. The Guild said it has seen an “uptick” of members complaining about potentially AI-generated books.
“Misappropriating authors’ names to sell scam books through Kindle and Goodreads is not new, but has gotten worse with the advent of AI-generated content,” Mary Rasenberger, the Author’s Guild CEO, wrote in a statement. “It can sometimes be difficult to get problems like this resolved through Amazon’s regular channels if it is not a blatant copyright infringement…especially if the legal basis is not clearly articulated. Amazon needs to make it easier for authors to get issues resolved. We will continue our conversations with them about advancing their efforts to keep up with the technology.”
Her incident raises broader questions that are making their way through courts about how AI uses copyrighted works to “learn.” Just this week, though, Google said to the Australian government that AI systems should be able to scrape the internet, while also adding that opt-outs should be supported. AI-generated content is currently not protected by US copyright laws.
Will companies have to get the author’s consent for their models to be trained on copyrighted work? Is there a way to verify what’s human work and not? This is only the first chapter in what’s sure to be a long, complicated saga for Amazon.