Photo by Steve Johnson on UnsplashEveryone’s falling into line.News Corp is the latest media giant to ink a deal with OpenAI to license its content. Joining Axel Springer, the AP, the Financial Times, Dotdash Meredith, Reddit, and others, News Corp will make its publications — including The Wall Street Journal, The Times of London, and The New York Post — available for OpenAI to ingest into its large language models (LLMs), which will then use that information to answer user queries.How that will look in practice is still to be determined. As I reported previously, OpenAI sees recent news stories differently from archives, and the Journal’s reporting on the deal has a couple of new and surprising details that sync with this approach.First is that the agreement includes a guarantee from OpenAI that newly published content “won’t become available on ChatGPT immediately after publication.” That particular note has a single source and is lower down in the piece, so take it with an extra shake of salt. But it’s something that hasn’t been mentioned in any of the reporting on previous agreements, and it suggests a couple of things:

This limitation also makes practical sense. As anyone who’s been on the internet more than five minutes knows, breaking news creates a lot of confusion. In those first few minutes or hours after a story breaks, rumors and exaggeration spread wildly, creating a fog around the truth. If there’s any part of journalism that absolutely requires human oversight, it’s sorting out the facts of a breaking story.That said, there’s no indication of exactly how long new content will be kept out of OpenAI’s clutches. A few hours? A day? A week? If it’s too long, the search function won’t be of much use for news, and competitors can already create generative summaries within a few hours of events. For instance, Perplexity’s digest newsletter includes summaries on fresh news stories, and they generally do a good job of encapsulating the main facts and analysis, like all the hubbub over Scarlett Johansson staring down Sam Altman over her voice.

The Chessboard
 The Journal article also revealed more about what the chessboard looks like in the ongoing game between the media industry and OpenAI. The cash OpenAI is paying is in the seven-to-eight-figure range, the piece said, with some pubs getting as low as a few million and some, Axel Springer in particular, garnering as much as $30 million.This probably has a lot to do with the size and value of the company’s content, in particular their archives, but it also shows an evolution to how OpenAI is thinking about these deals as competitors like Perplexity show that there is demand for a good generative search product.The problem that OpenAI faces, though, is that it has a target on its back. There’s a reason The New York Times is suing OpenAI and not other LLM creators — it has tremendously more influence than anyone else in the market. By making these deals, though, OpenAI turns this vulnerability into a strength: Not only does it preempt legal action; it also puts publishers in a position where they’re motivated to actively participate in creating a better generative news product.That’s an edge that could be decisive as generative search becomes the norm, according to Alex Fink, CEO of Otherweb, an AI-powered news aggregator. “They can now have a team from these publishers supporting them and making content easier for them to crawl, whereas all their competitors are forced to scrape the content adversarially,” he told me. “OpenAI isn’t buying the right to this content here; they are buying easy access and support.”

Google’s Rules
 Of course, there’s another major player waiting to play the winner of this chess match: Google. After it revealed its search generative experience (SGE), now called AI Overviews, would become the norm, Google is poised to exploit its position as the default search engine of the world to dominate search in the AI era.In contrast to OpenAI, Google has massive leverage with the media. It has made no attempt to make deals with publishers for AI Overviews, and why should it? Regardless of whether or not you see it as a fair extension of the web crawling it’s been doing for decades, no publisher in their right mind would block Google from indexing their content, since the entire media industry has grown dependent on referral traffic. We all know that traffic will drop even further as AI Overviews spread, but is anyone really willing to let it crash to zero overnight?As a publisher, it almost makes you want to root for OpenAI. At least with them, you have a chance of making a deal, if you’re big enough. But that depends on them making a better mousetrap —  a version of ChatGPT that returns better topical information, and paths to go deeper, than Google or anyone else.Checkmating Google on AI search is not going to be easy. But with the evergreen incentive of cold, hard cash, OpenAI is slowly getting the media on its side.More By This Author:ChatGPT Search? What It Would Mean For NewsCrowing for AI Regulation, OpenAI’s Search Surprise, and MoreWhat the Rise of Perplexity Represents