The New York Times has marshaled a major legal challenge against AI companies, alleging that Microsoft and OpenAI engaged in wide-scale copying, hijacking the Times' journalism to train its AI chatbots. The lawsuit is the latest in a broader dispute over how courts should view the legality of training large language models using the published work of others without compensation.
Permissive interpretations of AI and intellectual property could shield AI companies from liability not just when it comes to disputes with the Times and other news outlets, but with visual artists, record labels, and authors who have filed their own lawsuits against the makers of generative AI tools.
Sam Altman, left, CEO of OpenAI, appears onstage with Microsoft CEO Satya Nadella at OpenAI DevDay. (Barbara Ortutay/AP Photo) ·ASSOCIATED PRESS
The legal battles have wide-ranging implications because they have the power to shape the future of AI development, potentially restricting how companies use data on the web to build and improve their products, like the popular chatbot ChatGPT. These cases also carry huge importance for creative professionals who see the dawn of AI tools as an existential threat to their professions and yet another iteration of Big Tech exploiting the labor of others and reaping the rewards.
The Times contends that millions of its articles were used without permission to train AI products, which use the authority of its journalism to provide reliable information to users in ways that now compete with the Times.
"Defendants seek to free-ride on The Times's massive investment in its journalism by using it to build substitutive products without permission or payment," the lawsuit, filed in Federal District Court in Manhattan, said.
In a statement, OpenAI said, "We respect the rights of content creators and owners and are committed to working with them to ensure they benefit from AI technology and new revenue models."
Rick Allen, the co-founder of Nautilus Productions, a boutique stock footage company and production house, said generative AI models would not function without other people’s creative input.
"There are fundamental rules when it comes to copyright: No. 1, don't steal. And No. 2, if it's not yours, get permission to use it. That applies to me or Disney," said Allen, who also runs the Facebook group Artists Against Copyright Infringement.
"If you need this data and this information, you do what the rest of us do, you pay for it," he added. "There's an established way of doing business, and the AI companies have just ignored that."
AI backers have pushed back against this traditional view of copyright law. Venture capital firm Andreessen Horowitz, for instance, has said that "imposing the cost of actual or potential copyright liability on the creators of AI models will either kill or significantly hamper their development."
The VC firm, which has invested in OpenAI, went on to say that treating AI training as copyright infringement would lead to "far less competition, far less innovation and very likely the loss of the United States’ position as the leader in global AI development."
But critics say the tech world has sung this tune of unbridled innovation before.
•USD
(NYT)
"There’s a tech-utopian narrative that says stay out of the way, we've got this, we are changing the world, and any costs to us, whether it's legal liability or licensing demands, stand in the way of progress," said David Newhoff, an author and copyright advocate. "We have been told this many times before," he said, pointing to a litany of social problems that emerged alongside the rise of massive tech platforms.
Some financial analysts see the Times' dispute resolving itself in more amicable terms, in which AI companies get to keep using creative material to train their AI models by paying the institutions that produce the work. Other news outlets and publishers, including Axel Springer and the Associated Press, have already agreed to licensing deals with OpenAI.
In a new note from Evercore ISI, analysts said that investors have begun to view AI copyright disputes as a potential growth opportunity.
"We believe the most likely outcome is that NYT signs several AI licensing deals over the next few years that are each worth low tens of millions of dollars of revenue per year," the analysts said.
Times investors appear to be banking on that outcome. Shares of the newspaper have traded as much as 4% higher since news broke of the lawsuit.
The changing landscape of the web presents additional financial incentives for media companies considering agreements with AI companies.
As more users access information and content without using traditional web browsers, the advertising revenue tied to search will diminish. That poses a threat to companies reliant on such ad-based models. AI tools like ChatGPT, which answer users’ questions as if they were speaking to a person, are considered a more intuitive replacement to the formulaic search engines that have defined the internet experience for decades.
More than one-third of the Times’ web traffic comes from organic search, Evercore analysts noted, highlighting the risks of a drop in search-based revenue.
Even if the Times settles, advocates for creative professionals worry that the livelihoods of work-for-hire artists could be threatened if AI companies are free to hoover up content on the web while cutting deals with major institutional players.
"The New York Times is a very big fish in a small pond. On the other end is small creative professionals," said Allen. "We don't have the ability to fight or to stand up against AI misuses of our intellectual property."
Hamza Shaban is a reporter for Yahoo Finance covering markets and the economy. Follow Hamza on Twitter @hshaban.
