AI is a complicated form of machine learning. What does that mean? Who knows — we’re no experts — but it involves training a machine by having it read hundreds upon thousands of books, both fiction and non-fiction, which becomes a little problematic when you consider that none of the authors whose works were used to train software like Chat GPT were informed or compensated for their works being used in such a way.
It gets worse when you consider the possibility that these copyrighted works were obtained from shadow libraries and downloaded in bulk. Of course, this hasn’t gone over too well with authors in general who see the threat that AI is posing to their livelihoods. Novels written by robots are starting to appear on websites like Amazon and other stores and these products are claiming to be the work of well-known authors. People already hate AI for taking jobs in the creative field but it somehow always manages to get worse.
Are authors being stolen from?
It’s even worse for the authors whose works were allegedly stolen and used to train the AI. Not only has George R.R. Martin’s works been used to train the software, but fans have already used that software to write his sixth and seventh Game of Thrones books. Understandably, the much-loved author isn’t taking it lying down as he and 16 other writers are currently working on a class-action lawsuit against OpenAI.
The lawsuit is being led by the Authors Guild and includes other well-known names like crime thriller writer John Grisham and ‘My Sister’s Keeper’ author Jodi Picoult. The lawsuit was filed on Tuesday, Sep. 19, and seeks to stop OpenAI from using the named authors’ works to train the software. The lawsuit also claims for damages, although the specific amount is not stated.
“Without plaintiffs’ copyrighted works on which to “train” their LLMs, defendants would have no commercial product […] Defendants willful copying thus makes plaintiffs’ works into engines of their own destruction.”
The main concern is with the infringement of copyright, you can read the complaint for yourself here, but it argues that OpenAI could have used public domain works or paid a fair fee to authors for the use of their work. However, they did none of those things. Then, to rub salt in the wound, the software allows people to generate author’s works for a small price, or even nothing at all. Odds are that an AI-generated book from robot George R.R. Martin will be nothing like the author’s actual work, but it’s close enough to damage potential sales and his reputation.
This isn’t even the first lawsuit filed against OpenAI as the company, alongside Meta, is currently battling a similar suit from July this year involving author Sarah Silverman’s claims that the company illegally used her works to train the software.
OpenAI has not commented upon the legality of the acquisition of its datasets. These datasets contain a staggering amount of books and, in yet another lawsuit against the company, it’s stated, yet again, that these infringe on the author’s copyright.
So will the lawsuit be successful?
It’s complicated as this is an entirely new situation. However, one thing that does seem to be certain is the fact that the author’s copyright has been infringed and that their future profits have potentially been damaged by this new software. There is a compromise here, if OpenAI sticks to public domain works, or better yet, pays a fee to authors, then everyone can be a winner. Whether they’ll do that remains to be seen.