No, I’m not talking about the newest ways writers can use AI (just don’t, if you ask me), but rather on the latest developments for writers whose work has been or could be used for AI training, including for me. That was too long for a title, though, so apologies if you came here looking for some other news.
I recently signed my first-ever AI opt-in addendum with Bloomsbury for the eventuality that they negotiate a deal or deals to use A Writer’s Craft in AI training. I was a little surprised to get this, and didn’t have a lot of time to consider it, but fortunately, I’ve been aware of what’s going on enough that I could make my decision. This doesn’t mean that my textbook will be used for AI training, but if does mean that my publisher can negotiate subsidiary rights to do that. Likely, this would be one on behalf of all their titles or at least a subset of their titles that the AI company is interested in. Bloomsbury also made it clear that opting in would likely make your title more accessible to AI search and other uses of AI that would make it more visibile in the future.
But why would I opt in if I’m opposed to AI? I will say that I’m not 100% opposed to its use for things like research where it could be very helpful. I am opposed to creative writers relying on it, though. And I’m opposed to the current AI engines because they have been trained on databases of pirated texts. I don’t trust AI, but I also don’t think they should be making billions when they have essentially stolen the content they used to get where they are.
Licensing agreements are one way that authors can get paid for the use of their work. They are also a way for publishers to enforce certain restrictions about how the work can be used, so that it can’t be used to create a competing work, for instance. Because licensing could lead to some income for writers and because it can also lead to some protections for writers and publishers, I support it. The opt-in addendum only means that my book will be considered for this, but it’s a starting point.
The Author’s Guild has advocated for these kinds of agreements, and I can see why, collectively, they see them as the best legal means to protect authors’ rights going forward. They have also been involved in lawsuits with AI companies, and have recently reported a settlement agreement with Anthropic where each author whose work appeared in the pirated databases they used would be paid $3000 compensation per title.
Oddly enough, this could affect me as well. This summer, I was surprised to learn that my third poetry collection Barrier Island Suite was included in one of the pirated databases they used. I suppose I shouldn’t have been, since a Google alert I have on myself had periodically turned up links to pirated copies of the book. I initially alerted my publisher, though I don’t know that they ever did anything about it.
Whether I’ll ever see my $3000 remains to be seen (probably minus legal fees, etc, so it might end up buying me a sixpack or two if I get it). That will depend on whether it is determined that my title was one of those actually used. A while back I reported it as one that appears on a list of titles in the database, but I don’t know more than that. It also depends on whether my publisher ever registered the copyright with the U.S. Copyright Office, which I believe they were supposed to do, but some publishers rely on only registering the ISBN and listing a copyright notice in the book. So if Texas Review Press didn’t dot all its i’s and cross all its t’s, I could be out that money. Or my contract may specify that only a portion of subsidiary rights goes to me, and I might not see much, if anything.
Still, it’s gratifying to know that some writers are getting compensation from at least one company. And it’s likely more settlements will follow. $3000 per title may seem like a pittance, but with the sheer number of titles included in these databases, which could number in the millions, the payout would be astronomical. Not all titles will meet the criteria for being included, though. Books must have an ISBN or ASIN have been registered with the copyright office in time, and it’s estimated that about 500,000 will qualify. The total payout is reported to be $1.5 billion.
There are, of course, many other AI companies, but hopefully this settlement will set a precedent that authors and publishers need to be paid for the use of their work. I don’t expect to get rich off of selling AI training rights to my work, but I also don’t like to see companies get rich by stealing access to the words and ideas of human writers, especially when they have been stealing that access in order to train a technology that could make our words less valuable.
I personally don’t believe that project will be completely successful, since I believe in the value of human expression over AI-generated expression. But writers will be in competition with AI for the attention of readers, some of whom might be satisfied with the predicatable and formulaic output of a machine as opposed to the often messy, unpredictable writing the emanates from the human experience.
Kendall Dunkelberg 