Coinciding with the Hollywood screenwriters' guild strike, Game of Thrones author George R. R. Martin and other writers have sued California-based startup OpenAI for allegedly using their work to create Chat GPT without respecting their intellectual rights.

In their lawsuit filed Tuesday in federal court in New York, the writers accused the company of using their books "without permission" to train its language model, the artificial intelligence technology underpinning ChatGPT software, which is capable of producing all kinds of text by simply asking it a simple question.

The lawyers who filed the lawsuit on behalf of the book considered that "these algorithms involve systematic theft on a large scale."

The class-action lawsuit includes the authors' guild authors guild and a number of writers including George R. R. Martin and novelist John Grisham.

OpenAI and its competitors have filed several other lawsuits from artists, organizations and programmers.

Screenwriters strike

The screenwriters, who are part of a union of 11,500 film and television script writers, who have been striking in Los Angeles for months, have been demanding guarantees that artificial intelligence software will not be used either to write scripts for works or to reproduce their voices and images.

On the other hand, Hollywood studios recently suggested striking screenwriters to rework scripts created by the beginning of artificial intelligence programs, considering them the only author of this work, and therefore not receiving less pay.

On the other hand, the studios have not addressed the issue of training AI programs based on existing scripts, which the screenwriters' guild considers a red line, and an existential threat to their future careers as they fear that artificial intelligence could replace them soon.

The syndicate demanded that no automated production be given the status of "literary material" or "source", which are key terms related to copyright. It has also sought to prevent scenarios written by its members from being used to train AI programs.

Very few screenwriters believe that AI software may be able to do their job. But the mere interest of studios and streaming platforms in exploring these automated capabilities in this area is a threat and an insult, they say.

"Danger to a living"

In Tuesday's lawsuit, the lawyers argued that linguistic models "pose a risk to novelists' ability to earn a living, because they allow anyone to automatically generate, free of charge (or at a very low price) texts for which they are supposed to pay authors."

They cautioned that generative AI tools could be used to produce derivative content that imitates the style of the book.

They noted in the lawsuit that "the deliberate copying (of the plaintiffs' work) diverts their actions in an unfair and perverted manner (...) to engines to destroy them."

The writers and the syndicate are demanding damages and a ban on the use of copyrighted books "without explicit permission" in the training of language models.

Battles

OpenAI needed large amounts of online script to train its language model, but it didn't specify exactly the locations and scripts used.

The company, which has become an artificial intelligence giant thanks to the widespread success of ChatGPT, faces a number of other similar lawsuits, including those filed by a group of computer engineers who also claimed Microsoft, the startup's main investor, and GitHub.

Artists sued Stability AI, Midgorney and Developer Art, whose programmes were trained by a large number of online visual works.

In early September, Microsoft announced that it would provide legal protection to customers prosecuted for copyright infringement over content created using its generative AI tools.

Months ago, authors began filing lawsuits to protect their rights to counter artificial intelligence that uses their work to generate content, but their legal battles will not be easy: In Europe and North America, the law tends to endorse artificial intelligence, although the situation may change, according to jurists.

U.S. law authorizes data mining as part of a so-called "fair use policy," a lawsuit against Google linked to the digitization of books, won by the U.S. conglomerate, which owns the Internet's largest search engine.

For AI-generated content, the legal situation is difficult. Can this work be classified as counterfeiting, especially if the user of the artificial intelligence software requests a work that mimics the "style" of the author or imitates a certain logo?

In this context, French and European law, like American law, recognizes counterfeiting only in the case of copies of a specific act.