Did we write this descriptive text? Or was it generated by AI? You may never know.
Today’s artificial intelligence can turn your fun selfies into Renaissance portraits, change the appearance of actors on the big screen, or casually chat with you over Facebook. And it can also write entire novels from scratch.
As of today, free tools are available for anyone to feed a published text into a piece of software or AI language model and produce a new work. Before you dismiss a work generated by algorithms as unreadable—or for entertainment value only—already writers have been experimenting and doctoring the results to produce something market-viable. If such a work is then sold, who owns the copyright? What constitutes illegal or unfair use of an author’s work when it is consumed and reconstituted or mimicked by artificial intelligence?
One author is taking a Big Five publisher to court over possible software manipulation of her work, albeit in a decidedly low-tech manner. Author Pamela DuMond has filed suit against author Emma Chase and her publishers, alleging copyright infringement that she believes is the result of digital text spinning followed by a re-write and polish. Text spinning is a technique where an article—or an entire book—is fed into software that makes minor changes to create a seemingly original work. DuMond’s suit claims that too many similarities exist between her novel and Chase’s for it to be mere coincidence. Some character names are identical, and very unique words (such as Beyoncé) appear at roughly the same place in the story.
The publishers of Emma Chase’s novel include Diversion Publishing as well as Simon & Schuster, which released the audio edition. In January 2020, as The Passive Voice points out, the defendants’ attorneys on the case were replaced by Simon & Schuster attorneys, making this a very interesting case to watch.
This week, we talked with attorney Adam Losey, a business litigation lawyer who has experience with text-spinning cases. First, text spinning isn’t a new phenomenon; it’s been around since the 2000s and has been used by students to plagiarize papers. Even so, Losey says there is no bright-line rule you can apply that will tell you when spinning constitutes copyright infringement. It all depends on a court finding “substantial similarity” to the protected work. “The law is gray but intentionally gray,” Losey says. He points out the Supreme Court hasn’t taken on a lot of copyright cases, which has allowed for flexibility and evolving standards for substantial similarity, based on cultural norms. But on the flip side, no one has clarity on what crosses the line and what doesn’t.
Regardless, Losey says people are inclined to hate spinning for two reasons. One, it’s often been done to fool search engines (and people) into thinking that some material is new when it isn’t; and two, it’s considered lazy. Those who go to court over spinning must establish evidence that the work has been put through a “sausage grinder,” as Losey put it.
Spinning requires manual intervention and effort, at least in the case of novels, to produce a publishable product that might be successful, legal questions aside. But AI improves every year, and AI-generated short texts especially are becoming indistinguishable from human-crafted versions. Newspapers and magazines already use AI to generate publishable stories from scratch, with no infringement concerns. For a small taste of how the latest AI creative-writing generators work, try Google’s Poem Portraits or Talk to Transformer. (And you can read more about AI writing in this New Yorker feature by John Seabrook.)
Don’t underestimate how far the technology has already advanced. Last week, we attended a Digital Book World talk by Publishers Weekly’s Jason Boog, who has been researching how writers and publishers use AI in day-to-day creation. He said, “The most amazing thing I discovered: the tech is here already. … It’s like the early days of Photoshop for graphic designers right now for the publishing industry. We shouldn’t think of it as something we can’t use right now.” Boog pointed out that anyone can get access to the technology that powers AI writing tools, and it requires no special tech knowledge or coding. After setting up his own generator on his laptop at home and feeding the generator a series of sample texts (known as a dataset), Boog noticed right away that the AI was able to produce a hero’s journey.
When anyone can create a dataset—using multiple authors’ works or even just a single author—then use AI to generate a new work from scratch, who owns the copyright to the resulting work? These questions are so far unanswered. Boog says it’s analogous to the way someone can use a filter on Photoshop to immediately change the look and feel of a photo. You could create something that evokes a famous author’s work, but with an unusual—and possibly commercially successful—twist.
Taking that thought experiment a step further: Losey pointed us to a short story by science fiction writer (and attorney) Kevin Liu, “Booksavr,” published last year. The premise: a new software allows readers to change the sex, race, or orientation of the characters to make stories more inclusive. So, if you’re reading The Great Gatsby and you don’t think it is sufficiently diverse, you can spin it to change the identity of the characters. The story imagines the evolution of case law if such a software were indeed to exist. Losey says that current law might see a reader’s manipulation of a text to be fair use, but that once anyone moves to selling such a version, that’s different, and fair use is out the window.
Bottom line: In the text-spinning case, DuMond’s suit attempts to detail how the two novels are substantially similar, but it’s anyone’s guess as to whether that will meet the criteria required to win an infringement case. Losey mentioned to us that a copyright claim isn’t the only possible recourse an author might have with text spinners. Other claims—like a conversion claim or an unfair competition claim—might work. Since it’s not possible to protect an idea or a story trope, and it might be hard to prove substantial similarity, we wonder what type of new cases might arise in the age of AI, especially if copyright law doesn’t deliver the protection that authors or publishers desire.
Jane Friedman has spent her entire career working in the publishing industry, with a focus on business reporting and author education. Established in 2015, her newsletter The Bottom Line provides nuanced market intelligence to thousands of authors and industry professionals; in 2023, she was named Publishing Commentator of the Year by Digital Book World.
Jane’s expertise regularly features in major media outlets such as The New York Times, The Atlantic, NPR, The Today Show, Wired, The Guardian, Fox News, and BBC. Her book, The Business of Being a Writer, Second Edition (The University of Chicago Press), is used as a classroom text by many writing and publishing degree programs. She reaches thousands through speaking engagements and workshops at diverse venues worldwide, including NYU’s Advanced Publishing Institute, Frankfurt Book Fair, and numerous MFA programs.
