By Madison Overmier
AI training practices have been using authors’ work without consent for AI training, including pirated databases like Library Genesis (LibGen).
Library Genesis is a digital library that provides anyone with free access to copyright-protected content. Content on the website ranges from scholarly journal articles to magazines.
Controversy escalated on March 19, when court documents were released that said the CEO of Meta, Mark Zuckerberg, permitted staff to use LibGen to train AI model Llama 3.
On March 20, The Atlantic published a tool to search LibGen for material possibly being used to train AI programs, where many authors found their work.
Sarah Smith-Frigerio, assistant professor of public relations, recently co-edited a book that triggered her interest in the issue.
Smith-Frigerio said, to her knowledge, most large language models and image generators are being trained with copyright materials without permission from the original creators.
“I do not agree with using another’s words or creations without their permission or knowledge,” said Smith-Frigerio. “We need policy and law to catch up with generative AI.”
Piracy and copyright policy were created to protect original works of art. Authors can spend a large amount of money every month on scanning services that monitor sites taking their work. If their work is found, they have to send takedown requests through the Digital Millennium Copyright Act.
Smith-Frigerio said it’s time to seriously reconsider copyright policy and law.
“That doesn’t mean just making new laws, but perhaps even re-envisioning what intellectual property is and who gets to use things, as well as how,” said Smith-Frigerio.
With discussions evolving about copyright laws, organizations like the Tampa Writers Alliance (TWA), founded in 1989, help writers share their concerns and provide support to each other.
Writer Raven Snouffer is a member of TWA, which is a diverse group that has spent years supporting all writers on their journey to become better.
Snouffer said there are multiple independent authors whose books have been scrapped through LibGen. LibGen is a Meta used to train AI models.
“I think it’s bald-faced theft; simply put, I think none of it would work if things were used with permission, and I personally don’t know any others who would give permission,” said Snouffer.
The Authors Guild is another organization that supports authors. They successfully took down another piracy site, Z-Library, back in 2022. On March 20, they released an article with step-by-step instructions on what actions authors can take.
Snouffer said there is a lot of doom surrounding the topic but also a lot of hope and reassurance, which is powerful.
“One person I was talking to specifically had talked about a likeness to the resurgence of vinyl, so if this does take over, there will be some sort of analog grassroots community that can be built as a counterculture to this,” said Snouffer.
Melissa Saneholtz, an author who had her stories stolen from her, said the support from other authors in the writing community is the only bright spot in this.
Saneholtz said she was devastated when she found out the years she spent crafting her stories were stolen from her.
“Blood, sweat, and literal tears went into every single book,” said Saneholtz. “For someone with all the money and power in the world to swoop in and steal that? It’s wrong on so many levels.”
LibGen is said to be registered in Russia. Given that the site is outside of the United States, the elimination of these pirated materials remains difficult. The database contains over 81 million research papers and over 7.5 million books.
Saneholtz said you are not a genius if you rely on more creative, talented people to do the work for you.
“You’re not a visionary. You’re not someone to be admired. You’re lazy. You’re weak. You’re spineless,” said Saneholtz. “But they don’t care because money and power reign supreme with these people.”
Legal action has already begun against Meta and other AI companies. Kadrey v. Meta, a class action lawsuit, has automatically enrolled individuals whose books were used by Meta.
The court will first decide whether Meta broke copyright laws. The Authors Guild said a decision should be expected by this summer.
—
Photo courtesy of Getty – Igor Kutayev.