As artificial intelligence (AI) threatens the revenue of news publishers, Barry Diller—the billionaire founder of media conglomerate IAC—wants the industry to fight back.
News publishers should band together to take on large language models like ChatGPT and assert their copyrights in court, Diller said at the Semafor Media Summit yesterday (April 10).
While it isn’t clear where exactly these chatbots get the information to answer user questions, the assumption is that at least some of it comes from the articles published by news sites, which threatens their businesses.
“Many have speculated that GPT-4 (ChatGPT’s newest model) has more or less swallowed the Internet whole,” said Gary Marcus, who founded AI companies Robust.AI and Geometric.AI, via email.
Over the last 20 years, news publishers have seen their advertising revenue decimated by the growth of the internet. The U.S. has lost 2,500 local newspapers, a quarter of the industry, since 2005, according to a report by Northwestern University. Inflation and a downturn in the advertising economy have also contributed to financial uncertainty in recent years, with 360 newspapers closing since just before the Covid-19 pandemic began, Northwestern reported. But the most recent threat is AI.
The income that publishers make from advertising is compromised when original reporting appears in generative AI results, said Alex Connock, University of Oxford business school fellow, in an email. Connock also wrote a textbook about media business models in the age of AI.
Articles that appear on social media and search engines link back to the original publications, which allows them to collect advertising revenue when consumers click though for more information. At the very least, the publisher gets credit. But early models of AI chatbots, including ChatGPT, don’t automatically include links to the source of information it responds with. As the use of chatbots becomes more mainstream—potentially replacing the need for consumers to click into news sites—publishers could face significant revenue declines.
If publishers don’t find “some way of getting compensated for it, all will be lost,” Diller said.
Could news sites align against AI?
“The idea of publishers banding together is lovely, but it’s never really happened before,” said Charlie Beckett, director of a media think tank at the London School of Economics and Political Science. The nature of journalism is incredibly competitive, and these companies are like “cats fighting in a sack,” he said.
There’s no way for publishers to take on AI using technology, he said. While AI companies are at the forefront of technological development, publishers are using paywalls, a decade-old invention, to guard their site. The extent to which AI can read past paywalls isn’t clear, and paywalls aren’t always impenetrable. “Publishers are in a different century to these people,” he said. If news sites could gather to take on large language models, it would likely be through class action lawsuits, he said.
“Companies can absolutely sue under copyright law,” Diller said at the conference. The case would likely end up in the U.S. Supreme Court, he said, and the ruling could result in a precedent for how large language models interact with publications.
It is also possible that individual newsrooms make licensing deals with AI engines, said Beckett, which could be a huge pay day for publishers but exclude others.
