“If you think that won’t happen,” he added, “you’re just being a fool.”
This month, however, Diller’s IAC — which owns People, Food & Wine and InStyle, among other well-known magazines — signed a “strategic partnership and licensing agreement” deal with Microsoft and OpenAI, the creator of ChatGPT. OpenAI will now be allowed to tap into some of the company’s copious archives, in exchange for providing links (and, theoretically, web traffic) back to the original stories.
As tech companies race to perfect machines that can already produce humanlike text, summarize long documents and describe images and videos, media companies are struggling to figure out where they fit into the new gold rush.
And while some are continuing to wage a war — notably, the New York Times, which filed a copyright lawsuit against OpenAI last year — others are trying to find some uneasy accommodation with the very technology that some fear is poised to destroy them.
Diller said he sees “no contradiction” between last year’s battle cry and this year’s truce. IAC, he said in a statement, will get “direct compensation for our content” (the financial terms have not been made public) while “continuing in whatever forum to enforce copyright laws against others.”
He was hardly the first media boss to find room for compromise. The industry’s warming to AI began last summer, when the Associated Press reached a licensing deal: OpenAI would get access to part of its vast trove of news stories, while the AP will “leverage OpenAI’s technology and product expertise,” according to the announcement.
Loose chatter of media companies putting up some kind of united front against Big AI fizzled away. Deals were announced with quiet efficiency.
Semafor launched an AI-generated breaking news feed “supported” by Microsoft. Axel Springer — the German publisher that owns Politico and Business Insider — said it would “leverage AI for enhancing content experiences” and “support a sustainable future for journalism.” Le Monde in France and Prisa Media in Spain partnered with OpenAI to bring French and Spanish news content to ChatGPT. When the Financial Times announced its deal with OpenAI in April, CEO John Ridding struck a Promethean tone.
“As with any transformative technology, there is potential for significant advancements and major challenges,” he said, “but what’s never possible is turning back time.”
OpenAI has refused to say what data goes into its technology. But all AI chatbots are trained on trillions of sentences scraped from the internet that somebody wrote. A 2023 review by The Washington Post of one important data set used by Google and Meta to train their AI found that it contained Wikipedia pages, personal blog posts and religious websites, as well as massive amounts of news articles from sites as varied as the New York Times and Breitbart News.
But now that OpenAI and other chatbot companies are reaching maturity as consumer products, they need more than just language-training fodder. They need up-to-date information to answer users’ questions about the world and current events.
In a first line of defense, many wary news companies started blocking OpenAI from scraping their sites. So the tech company started working on deals to pay for access to news.
Last week, News Corp. announced a multiyear deal that will allow OpenAI to tap into its news content to answer users’ questions. The Wall Street Journal — a News Corp. property — reported the deal could be worth more than $250 million over five years.
It’s a big number. But compared with “the value that generative AI is bringing to these large tech companies, I think it’s safe to say the amount that is being paid out is on the lower side,” economist Haaris Mateen told The Post.
Though AI companies cast a wide net for fresh content, journalism has turned out to be one of their most valued resources, added Mateen, an assistant professor of finance at the University of Houston. Otherwise, “they cannot produce results on current affairs — and when I say current affairs, I mean everything from politics to entertainment.”
But it costs money to hire the reporters and editors who create journalism. What’s shaping up, Mateen said, is a system in which “the costs of producing news is being borne by one side, while all the value is poised to be reaped by the other side.”
Some observers think the media companies rushing to partner with AI will come to regret it.
“Licensing deals to train the Automation Death Star to be able to more precisely replicate your work in the future,” journalist Hamilton Nolan wrote last week, “is the equivalent of feeling pleased with yourself that you made five bucks selling your house keys to some burglars.”
The New York Times is seemingly holding tight to its house keys. In December, it sued OpenAI and Microsoft for copyright infringement, arguing that the companies had used millions of Times articles to train the large language models they are now deploying to compete with the Times for readers and web traffic. The Times is claiming “billions of dollars” in damages.
In the papers filed in federal court, the Times argued that OpenAI threatened its very ability to make money from its own product.
“With less revenue, news organizations will have fewer journalists able to dedicate time and resources to important, in-depth stories, which creates a risk that those stories will go untold,” the Times asserted. “The cost to society will be enormous.”
OpenAI has said that its use of copyright articles is protected under “fair use” statutes.
But other media organizations are now adopting the Times’ stance. In February, the Intercept also sued OpenAI for copyright infringement. Last month, eight local and regional newspapers owned by Alden Global Capital sued for copyright infringement as well as reputational damage they claim was caused by OpenAI concocting nonsensical or “hallucinated” answers and attributing them to the newspapers’ reporting.
The fact that so many media companies are cutting deals with Open AI could “dilute” the leverage that the companies suing it have, Mateen noted. On the other hand, by paying some publishers so much money, Open AI may be undermining its own defense: If it were truly “fair use,” he said, “they’d be confident enough not to pay anything.”
Some other major publishers have neither partnered with OpenAI nor sued it — yet.
The Post and the national newspaper chain Gannett, for example, have opted to develop their own AI tools. Last week, both companies unveiled AI-generated, newsroom-reviewed summaries appended to articles, a move they are billing as a way to engage readers more deeply with the articles themselves.
Post CEO and publisher William Lewis signaled that he is keeping options open. He said in a statement that The Post is “in the market for significant AI partnerships” while also referring to broader copyright concerns.
“We’ve got to get paid for what’s been taken so far, one way or another,” he said.
OpenAI CEO Sam Altman has described the licensing contracts as beneficial both to his technology and the quality of journalism. The CEO of News Corp. called his own deal with OpenAI the “beginning of a beautiful friendship.”
But some publishers hear echoes of their industry’s excitement over past waves of digital-era innovations — such as the promise of social media, which lulled media bosses into letting Big Tech supplant their relationships with their audience, while getting little in return, according to Jessica Lessin, founder and CEO of the Information, a tech news site.
“For as long as I have reported on internet companies, I have watched news leaders try to bend their businesses to the will of Apple, Google, Meta, and more …” Lessin wrote in the Atlantic. “It never, ever works as planned.”
