Two lawsuits for copyright infringement filed by authors against OpenAI, an artificial intelligence company, have been partly dismissed in a court of law.

The cases, led by comedian Sarah Silverman and novelist Paul Tremblay, alleged that OpenAI unlawfully used their books to train the large language model underlying artificial intelligence tool ChatGPT, which produces human-like text in response to prompts.

A federal judge in California ruled on Monday to dismiss the majority of the writers’ claims in response to OpenAI’s motion.

Judge Araceli Martínez-Olguín of the district court also stated that the lawsuits would be combined with a related complaint filed by a different set of writers, such as Michael Chabon, Ta-Nehisi Coates, and Andrew Sean Greer.

According to Martínez-Olguín, a lawsuit alleging vicarious copyright infringement was rejected due to the authors’ failure to demonstrate significant similarities between their books and ChatGPT’s generated content. Additionally, she stated that the authors’ argument that all of ChatGPT’s outputs constitute infringing derivative works was deemed inadequate.

Some other accusations that were rejected were claims of carelessness, gaining unjustly, and breaking the Digital Millennium Copyright Act.

Despite this, OpenAI still faces accusations of violating unfair competition laws through its use of copyrighted books without the permission of the authors.

The decision is similar to a previous case where Silverman sued Meta for using protected books in their AI training tool, LLaMA. In November, the judge mostly ruled in favor of Meta, but the accusation of direct copyright infringement will continue to be investigated.

While Silverman had accused OpenAI of direct infringement in the lawsuit, this accusation was not included in the company’s request for dismissal. However, Martínez-Olguín stated that vicarious infringement “requires a demonstration of direct infringement,” which the authors were unable to provide.

The group of authors who are suing OpenAI in California recently requested that Martínez-Olguín cease a similar lawsuit that was filed in New York by the Authors Guild and various novelists, such as Jonathan Franzen, Jodi Picoult, David Baldacci, and George RR Martin. The lawsuit alleges that OpenAI is engaging in “forum shopping” to secure a more advantageous schedule.

In August, the Atlantic reported that over 170,000 books written by various authors such as Zadie Smith, Stephen King, Rachel Cusk, and Elena Ferrante were utilized to educate Meta’s LLaMA and potentially other generative-AI tools.

In the June lawsuit filed on behalf of Tremblay and Mona Awad – who has since withdrawn from the suit – the authors’ attorney Joseph Saveri wrote that one of the “internet-based books corpora” that OpenAI said it used to train ChatGPT-3 is estimated to contain nearly 300,000 titles, and that the only websites to offer that much material are “shadow libraries” such as Library Genesis (LibGen), through which books can be secured in bulk via torrent systems.

Martínez-Olguín granted authors an extension until March 13th to revise their complaint.

Source: theguardian.com