This summer saw a string of authors suing ChatGPT’s owner, OpenAI. Last week, the Authors’ Guild filed a class action against the tech giant. What makes this lawsuit particularly significant is the number of high-profile authors whose names are attached. The name most of the headlines focus their attention on is Song of Ice and Fire author George R R Martin. ALLi's Self-publishing News reports on this latest lawsuit against OpenAI.
The lawsuit is ostensibly an accusation of copyright violation. It claims that that OpenAI used pirated shadow libraries to train its large language model. Such libraries do, of course, violate the copyright of the authors whose works they contain. Just last week, I reported on Library Genesis, a shadow library which is subject to a lawsuit from four large academic publishers, and has already lost a similar $15m suit to Elsevier. It is less clear what, if anything, those who use shadow libraries are violating in a legal sense.
You might be left wondering whether, if ChatGPT really has been fed Martin’s works, it might help the author to actually finish Winds of Winter. Well, it turns out one of the examples cited in the complaint is that fans have been able to use ChatGPT to generate versions of the famously delayed last book in the Song of Ice and Fire series. It claims this would be impossible without the model having been trained on works including the existing books.
What is abundantly clear is that this is about more than copyright. It is about an industry trying to protect itself from what it perceives as an existential threat. The concerns are the same as those of the Hollywood screen writers currently on strike. Those writers want assurances that AI will not replace them in the industry’s writing rooms.