This week, two companies involved in artificial intelligence development emerged victorious in a copyright lawsuit brought by a book author.
In San Francisco, two federal judges determined that human and Meta could utilize the book for training AI systems without needing permission.
On Wednesday, US District Judge Vince Chhabria sided with Meta, stating that the author failed to demonstrate adequately that Meta’s AI would harm the market in a way that qualifies as copyright infringement.
“The ruling doesn’t imply that Meta is using copyrighted materials to develop language models,” Chhabria noted, as reported by Reuters. “Instead, it indicates that the plaintiffs made a flawed argument and couldn’t establish a solid record for their claims.”
Amazon’s CEO announced that the implementation of AI has led to reductions in the company’s workforce.
The ruling from Judge Alsp came on Monday, referencing “Fair Use” by the authors Andrea Burtz, Charles Graeber, and Kirk Wallace Johnson, which humanity used to train its language model.
Though Alsp showed some support for the author, he mentioned that humanity had copied over 7 million pirated copies stored in what is termed the “Central Library,” which infringed the author’s copyright. He ordered a trial in December to assess the extent of this infringement.
Fair use serves as a crucial legal defense for tech companies, and the Allsup ruling marks a significant decision regarding AI generation.
AI firms argue that requiring them to utilize copyrighted materials properly and compensate copyright holders could stifle the rapidly growing AI industry.
Humanity and other leading AI companies, such as OpenAI and Meta, have faced accusations of downloading massive quantities of pirated books for system training.
Under US copyright law, deliberate infringement can lead to statutory damages as high as $150,000 for each work.
Copyright owners assert that AI companies are unlawfully replicating their content, generating competing products that threaten their livelihoods.
Chhabria expressed understanding of these concerns during a hearing in May, reiterating them on Wednesday.
He mentioned that generative AI has the potential to inundate the market with an abundance of images, songs, articles, and books, all created with minimal time and effort.
“By training generative AI models with copyrighted materials, companies can create outputs that significantly undermine the market for those works, hence diminishing the motivation for humans to create in the traditional sense,” Chhabria remarked.
