Artificial intelligence doesn’t just pop into the world fully formed.
In many instances, work from flesh and blood authors is pirated by large language models and used in the training of AI.
Author Aoife Barry said she found her book available on one such language model called Library Genesis, otherwise known as LibGen.
“I did a search, and I couldn’t find it at first and then when I posted about it on Bluesky, somebody responded to me and said, ‘Actually, there’s another search on the site that you can check,’” she told Moncrieff.
“I checked there, and the book was available there, as is, I mean, name any Irish author from the last at least 10-years and you’ll find their books or multiple copies of their books on LibGen.
“So, it emerged, according to the Atlantic and several court papers that you can find online, that it’s being alleged that LibGen is being used by Meta – which of course owns Facebook and Whatsapp and Instagram – to train what’s called ‘Llama 3’.”
Llama 3 is a large language model that competes with ChatGPT, which was created by OpenAI.

Ms Barry said this type of use of creative material has been “very upsetting for a lot of authors”.
“Unfortunately, we live in a time where I felt it was almost inevitable that my book could potentially end up on one of these things,” she said.
“Not because it’s a world-famous book, like, ‘Social Capital’ is an Irish nonfiction book, ironically about the tech world and social media – there's a heavy irony there.
“But it feels like any book that’s out there can end up potentially being pirated, because let’s not forget that you can buy my book in an e-book format.
“So, it’s easy enough for people who know what to do to kind of take an e-book and then to pass it along wherever they want to.”
'Not going to get very far'
According to Ms Barry, there is not much authors can do if they want their work removed from LibGen.
“There’s a court case obviously in the US, I think that’s a class action suit,” she said.
“From what I was reading, technically, if you’re an author and your book is on there, you’re part of that class action suit.
“But I mean, if you’re living in Ireland, I don’t know legally how that works – we don’t have class action suits over here, they do in the US.
“You can email LibGen and ask them to take your book off; that’s not really going to get you very far and nobody knows who runs the site.”
Ms Barry said that some publishers are striking deals with AI developers that would mean authors would have to be asked permission for their works to be used in AI training.
Main image: Woman writing in notebook and reading messages. Image: Panther Media GmbH / Alamy. 18 April 2022