A group of writers filed a copyright infringement complaint against Facebook, alleging that the business had stolen their works to train its artificial intelligence system, but a federal judge backed with Facebook parent Meta Platforms and dismissed the case.
U.S. District Judge Vince Chhabria’s decision on Wednesday was the second in a week that a federal court in San Francisco has dismissed significant copyright claims made by book authors against the quickly evolving AI business.
13 authors sued Meta, but Chhabria determined that their arguments were flawed and dismissed the case. However, the judge added that the decision only applies to the case’s authors and does not imply that Meta’s use of copyrighted content is acceptable.
Chhabria stated that this decision does not support the idea that it is legal for Meta to train its language models using copyrighted content. It merely supports the claim that these plaintiffs failed to create a record supporting the correct argument and instead made the incorrect ones.
“The court ruled that AI companies that feed copyright-protected works into their models without getting permission from the copyright holders or paying for them are generally violating the law,” said a statement from the plaintiffs’ lawyers, a group of well-known writers that includes comedian Sarah Silverman and authors Jacqueline Woodson and Ta-Nehisi Coates. However, the court decided in favor of Meta despite the undeniable evidence of its historically exceptional copyright pirating. To that end, we respectfully disagree.
Meta expressed gratitude for the choice.
“Fair use of copyright material is a vital legal framework for building this transformative technology,” the Menlo Park, California-based business said in a statement. “Open-source AI models are powering transformative innovations, productivity, and creativity for individuals and companies.”
Even though Meta’s motion to have the case dismissed was granted, it might be a pointless win. As Meta and other AI companies train their technology on human-written books and other works, Chhabria repeatedly cited reasons to believe that they have become serial copyright violators in his 40-page ruling. It also appeared as though he was inviting other authors to submit cases to his court that would allow them to go to trial.
The judge dismissed claims that forcing AI firms to follow antiquated copyright regulations would impede the development of a vital technology at a critical juncture. It is anticipated that the businesses creating these items would make billions or possibly trillions of dollars. The corporations will find a mechanism to pay copyright holders if they believe that training the models requires the use of copyrighted works.
U.S. District Judge William Alsup ruled Monday from the same courthouse that the AI company Anthropic did not violate the law by using millions of copyrighted books to train its chatbot Claude. However, the company still faces legal action for illegally obtaining those books from pirate websites rather than purchasing them.
However, according to Alsup, the actual process by which an AI system extracted text from thousands of written works to create its own text passages was considered fair use under U.S. copyright law since it was fundamentally transformative.
By stealing their novels from online archives of pirated works and feeding them into Meta’s flagship generative AI system Llama, the authors in the Meta lawsuit had claimed in court filings that Meta was culpable for widespread copyright infringement.
Generative AI chatbots may learn the patterns of human language from lengthy and uniquely produced text passages, like those found in books. The writers’ lawyers contended that Meta could and ought to have paid to purchase and license those literary works.
In court filings, Meta retorted that the new, AI-generated expression that emerges from its chatbots is essentially different from the books it was trained on, and that U.S. copyright law permits the unauthorized copying of a work to turn it into something new.
“After almost two years of litigation, Meta’s lawyers contended that there is still no proof that anyone has ever utilized Llama to read Plaintiffs’ books or even that they could.
According to Meta, even when requested to do so, Llama refuses to display the real works it has copied.
Llama’s lawyers wrote that no one could read Junot Diaz’s account of a Dominican lad growing up in New Jersey or Sarah Silverman’s account of her early years.
Accused of removing those books from online shadow libraries, Meta has further maintained that the tactics it employed are irrelevant to the nature and intent of its use and that the outcome would have been the same had the firm made an agreement with actual libraries.
Google established its online library of over 20 million books through these agreements, but it also faced ten years of legal challenges before the U.S. Supreme Court upheld lower court decisions that denied copyright infringement charges in 2016.
The writers’ case against Meta resulted in the deposition of CEO Mark Zuckerberg and revealed internal discussions on the morality of accessing datasets that have been illegally obtained for a long time.
In a court petition, the writers’ lawyers contended that authorities frequently shut down their domains and even bring charges against the offenders. It’s undeniable that Meta realized that stealing copyrighted content from illegal databases may put the business at serious risk, which is why it escalated the matter to Mark Zuckerberg and senior Meta executives for permission. Their risk shouldn’t be profitable.
Richard Kadrey, Andrew Sean Greer, Rachel Louise Snyder, David Henry Hwang, Ta-Nehisi Coates, Laura Lippman, Matthew Klam, Junot Diaz, Sarah Silverman, Lysa TerKeurst, Christopher Golden, Christopher Farnsworth, and Jacqueline Woodson are the plaintiffs named.
In the verdict, Chhabria stated that although he was forced to grant Meta’s summary judgment dismissing the lawsuit, the ruling’s overall effects were minimal. The verdict just impacts the rights of these 13 authors, not the innumerable others whose works Meta utilized to train its algorithms, because this is not a class action.
This story was written by Michael Liedtke, an AP Technology Writer.
