Will OpenAI’s deals with media companies lead journalists astray?

News organisations and publishers are facing a tough choice — they can either partner with AI companies or risk having their content taken over by AI companies in other ways.

The Murdoch family's multinational mass media company News Corp is the latest in a row of dominoes to strike a deal with OpenAI, allowing OpenAI to show its news content in response to queries users pose in ChatGPT.

The agreement means the makers of ChatGPT and other artificial intelligence tools, such as Sora, an AI model that can create realistic and imaginative video from text instructions, will have access to current and archived content from News Corp's major publications.

What could sharing journalism with AI companies and tech giants bring about for the future of news?

OpenAI's recent partnership with another major player in the world of media "certainly ensures that News Corp generates revenue streams from the use of its content to train AI models, but the amount OpenAI is paying under this deal is not high relative to the value of the data," professor of digital communication and culture at The University of Sydney and an Australian Research Council (ARC) Laureate Fellow, Terry Flew, tells TRT World.

"For OpenAI and others in this space, such as Google, it points to the problem they have been facing in acquiring credible data legally, and the need for a more mature business model."

Both companies involved did not disclose the financial details of OpenAI's latest deal, but according to the Wall Street Journal, which is owned by News Corp, the deal could potentially exceed $250 million over a five-year period and includes compensation in cash and credits for the use of the AI company's technology.

News Corp also owns MarketWatch and the New York Post in the US. In Australia and the UK, it owns The Daily Telegraph, news.com.au, and The Australian with the former, and The Sun, The Sunday Times, and The Times with the latter.

The multi-year deal announced last week follows the AI company's April 29 agreement with the London-based Financial Times to licence its content for AI learning development. Other publishers, such as Politico's parent company Axel Springer, The Associated Press, Prisa Media in Spain, and Le Monde in France, have also made similar deals.

"News is pure gold for an AI company," Jonathan Soma, a professor who teaches data journalism at Columbia Journalism School, tells TRT World. "It's written by humans, it's more recent than the knowledge cutoff of their chatbots, and it's what many of their users want to know about. If news organisations are going to sell the rights to such a precious substance, they should be setting prices sky-high."

Soma doesn't think AI will replace journalism; rather, it "will be a parasite latched onto the side of journalism."

"Without journalists, current-events chatbots would have nothing to write about, nothing to publish, no answers to give," Soma offers.

"The question is whether these partnerships can provide revenue to news organisations in line with the amount of effort required to produce the original journalism. I doubt that News Corp and other news organisations know how valuable their content is, and I doubt that OpenAI wants to pay what the content is actually worth."

Several companies, like the New York Times and the Chicago Tribune, have gone the other route and sued OpenAI and Microsoft for using their articles to train AI.

AI tools like ChatGPT, as well as Microsotf's Copilot and Google's Gemini, use large language models (LLMs) that analyse huge amounts of internet text to predict the next word in a sentence, allowing them to mimic human speech and writing.

Marius Dragomir, director of the Media and Journalism Research Center, an independent media research and policy think tank, highlights the importance of regulation for publishers collaborating with AI companies. While such partnerships may bring commercial benefits, not all content meets high standards of quality and accuracy.

Dragomir tells TRT World that it is crucial to recognise that significant disparities characterise the media industry, which could lead to further biassed coverage due to some media markets that "serve the interests of their owners." Some are heavily concentrated with power, while others are controlled by propaganda and disinformation, often propagated by government-funded media, he adds.

"In the case of News Corp, some of its media are known for their tabloid, not always accurate, low-brow content, so the questions arise: is this content well placed to be presented as the 'right' answers to whatever questions people might ask the AI?," Dragomir remarks.

According to Columbia Journalism School professor Soma, chatbots are also prone to hallucination and editorialising.

"It's easy for an AI to cite an article but give a completely incorrect summary – see the recent fiasco with Google's generative search results. These AI results, which have no concept of truth, could be a breeding ground for misinformation," he says.

For Pete Pachal, the founder of Media CoPilot, a Substack newsletter about how AI is changing media and journalism, OpenAI and News Corp's partnership "feels like a turning point."

While the future of news looked unclear when The New York Times sued OpenAI for copyright infringement in December, Pachal tells TRT World, "It feels like content deals will be the norm going forward," in light of agreements made with notable publishers such as the AP and Axel Springer—and most recently, News Corp.

"The New York Times has found a few minor allies in its lawsuit, but not many. That said, a ruling in the Times' favour, even partially, could change the entire picture."

"Tactically, the deals make sense since this will give publishers some badly needed revenue in the short term," says Pachal. Still, it may turn out to be a mistake in the long term since AI companies are essentially allowed to "own the customer relationship for AI summaries," he adds.

According to Pachal, Google is a "big player influencing the urgency of all this." He says publishers will suffer reduced traffic as the tech company's AI Overviews feature, which includes AI-generated summaries at the top of its search engine, becomes the norm.

"There will be no deals made with Google over content," Pachal explains, "since Google, rightly or wrongly, simply sees AI Overviews as an extension of the web crawling it's been doing for decades at this point."

He adds, "The biggest consequence of these deals with OpenAI is that publishers will be incentivised to root for OpenAI in any battle over the future of AI search. Because if Google wins, all that cash will dry up fast."

As average users become more comfortable with AI tools that directly respond to people's queries, it raises concern that people might rely more and more on major tech chatbots for information rather than reporting done by journalists and media companies.

Alfred Hermida, a professor at the University of British Columbia School of Journalism, Writing, and Media, where he served as director for over five years, explains to TRT World that news websites could potentially lose out on casual visitors seeking quick news updates, resulting in a significant decline in traffic when this happens.

"They are a source of valuable revenue for publishers, but could potentially harm them in the long-term as AI systems will be able to be trained on journalism content," Hermida says, adding that deals such as the one made with News Corp are important for OpenAI because it gives the AI company what it needs — a continuous supply of training material.

"What is less clear is what happens to the news media if and when systems like ChatGPT learn to produce good enough news reports."

The partnership could expand News Corp's audience to include people who don't typically visit its news outlets. However, Hermida says that whether this will boost readership for its brands remains unclear.

"The evidence from social media suggests that it doesn't help to develop brand loyalty as people identify the intermediary, for example, Facebook, as the source of the news," according to Hermida, who views deals between news organisations and AI companies as "a double-edged sword."

"The risk is that news publishers become dependent on AI companies, as they did on Facebook for visits and revenue."

The quality of these summaries and the overall user experience of AI-generated news queries will determine whether or not they become a primary source of information for people.

The University of Amsterdam's Erik Borra from the humanities and media studies faculty notes that people may need to get ready for a transformation in how news content is accessed and consumed.

As he puts it, "Users likely won't click through to the publishers. For publishers, it will thus become harder to 'track' what news is popular, and they won't be able to run ads next to it anymore (where generally more eyeballs is more money).

"It might also mean a decline in individual subscriptions."