A number of high-profile lawsuits in a New York federal court are threatening to derail ChatGPT and other AI systems that use large quantities of copyrighted human works
Could chatbots made of AI infringe on copyright and fair competition rules, though? This is especially true for the heavily advertised solutions from OpenAI and Microsoft. Professional writers and media organizations will have a tough time winning that argument in court.
My heart aches if I could cheer up the authors, but I just can’t. According to copyright attorney Ashima Aggarwal, a former employee of academic publishing giant John Wiley & Sons, the plaintiffs are in for a tough fight.
An example of this would be a lawsuit brought by The New York Times. Continuing the tradition of a small but distinguished collection of works by such well-known writers as John Grisham, Jodi Picoult, and George R.R. Martin. A third from bestselling nonfiction writers, one of whom penned the Pulitzer Prize–winning biography upon which “Oppenheimer” was based.
Though the particular charges differ in each instance, they all center on OpenAI, a San Francisco-based firm that is “building this product on the back of other peoples’ intellectual property,” as stated by Justin Nelson, an attorney representing both the nonfiction authors and The Times.
“What OpenAI is saying is that they have a free ride to take anybody else’s intellectual property really since the dawn of time,” Nelson added, referring to everything that is online.
In December, the Times launched a complaint against ChatGPT and Microsoft’s Copilot, claiming that these artificial intelligence systems are diminishing the influence of the news organizations they were taught to resemble and diverting attention and readers from copyright holders, such as newspapers and other media outlets that depend on advertising income to support their work. The fact that the chatbots were really rephrasing NYT articles was also shown. The newspaper said that its reputation took a hit because the chatbots would sometimes mistakenly associate false information with the publication.
With the addition of a fourth writer last week, two more nonfiction writers and a senior federal judge are now hearing all three cases together. U.S. District Judge Sidney H. Stein has been the head of the Manhattan court since 1995, when he was appointed by Clinton.
The New York cases have not yet been officially rebutted by Microsoft or OpenAI. However, OpenAI issued a statement this week stating that the chatbot’s ability to quote some articles was an isolated incidence and that the complaint against The Times was “without merit.”
“Fair use” is supported by long-standing and broadly acknowledged standards for training AI models utilizing publicly available web materials, according to the business’s Monday blog post. Rumor has it that the Times “either instructed the model to regurgitate or cherry-picked their examples from many attempts,” as stated in the story.
In its pursuit of a strong news ecosystem, OpenAI cited the licensing agreements it made last year with organizations like The Associated Press and the German media giant Axel Springer. For an undisclosed amount, OpenAI has licensed the news archive of AP. The New York Times followed suit after similar conversations.
Early this year, OpenAI said that its AI systems would be enhanced by access to AP’s “high-quality, factual text archive.” “Any single data source — including The New York Times — is not significant for the model’s intended learning.” Using a “enormous aggregate of human knowledge” to train its large language models, the business said in a blog post this week that news content is irrelevant to AI training.
In its arguments over copyright laws in the US, the AI business heavily relies on the “fair use” idea, which allows for certain limited uses of copyrighted material like research, education, or adaptation.
When asked about OpenAI and Microsoft’s behavior, the Times’ legal team answered on Tuesday by saying that it was “not fair use by any measure” as the companies were using money from the newspaper’s content investments “to build substitutive products without permission or payment.”
There has been a clear trend in court decisions favoring tech companies when it comes to applying copyright laws to AI systems. Visual artists suffered a defeat last year when the first big lawsuit against AI image-generators was partially dismissed by a federal court in San Francisco. Another California court has dismissed comedian Sarah Silverman’s allegations that Facebook parent company Meta infringed on her autobiographical material in order to develop its artificial intelligence model.
Despite the fact that more recent instances have shown more tangible evidence of harm, Aggarwal said that courts “just don’t seem inclined to find that to be copyright infringement” in relation to teaching AI systems to provide a “small portion of that to users.”
When faced with legal challenges, many IT companies look to Google’s defense of its online book collection as a model to follow. Google digitized millions of books and publicly displayed portions from them in 2016, and authors attempted to charge copyright infringement, but the U.S. Supreme Court maintained lower court judgments that dismissed their claims.
Judges evaluate fair use arguments individually, and their interpretation is “actually very fact-dependent,” says Cathy Wolfe, a board member of the Copyright Clearance Center—a Dutch firm that helps with license negotiations for digital and print media—and an executive at Wolters Kluwer. Many things have contributed to this, one of which is the effect on the economy.
“Just because something is free on the internet, on a website, doesn’t mean you can copy it and email it, let alone use it to conduct commercial business,” he added. Although I am unsure of the outcome, I am firmly in support of copyright protection for all parties involved. It encourages originality.
A number of news outlets and content creators are calling on Congress or the US Copyright Office to strengthen copyright protections for the AI era outside of the court system. Wednesday, a subcommittee of the US Senate Judiciary Committee will hear testimony from media business officials and others worried about how AI could affect journalism.
“Generative AI companies are using our stolen intellectual property to build tools of replacement,” CEO Roger Lynch of the Conde Nast magazine network plans to alert Congress.
“We believe that a legislative fix can be simple — clarifying that the use of copyrighted content in conjunction with commercial Gen AI is not fair use and requires a license,” a transcript of Lynch’s speech is read.
