Image: Getty Images
Imagine a meteor hurtling towards your city. It could obliterate you. But if you could somehow harvest its extraordinary energy, it might power your civilisation for centuries. The media industry now faces its own version of this cosmic dilemma. Artificial intelligence poses the most existential threat to its business model since the invention of the printing press. It also offers the only plausible path to salvation. Publishers must stop treating AI as just another technological shift to weather. They must learn to capture its power—or be destroyed by it.
The threat is shockingly clear-cut. The first digital age unpicked the industry’s economics by unbundling advertising from content. The AI age now promises to unbundle trust from information. When a reader asks a chatbot for the latest news from Ukraine, say, they get a summary—but no link to the reporting that produced it. The article itself, and the publisher who paid for it, become invisible. The value of a single story plunges to near zero. At the same time, the cost of generating plausible text, images and video is collapsing. The internet will soon be flooded with what insiders grimly call “AI slop”: synthetic news, commentary and analysis, produced at virtually no cost by machines. When everything is available for almost nothing, how can professional journalism survive?
The answer is found in the one thing the machines cannot replicate: trust. A chatbot can produce a coherent article, but it cannot stand behind it. It has no reputation to defend, no sources to protect, no ethical compass beyond its programming. In economic terms, AI risks creating a “pooling equilibrium” where readers cannot distinguish reliable reporting from algorithmic fabrication. In such a world, as the Nobel laureate George Akerlof showed in his work on “lemons”, the bad drives out the good. If you cannot tell an AI fake from a human-written fact, you eventually assume everything is fake—and you stop paying for any of it.
For media firms, therefore, the task is to prove their work is not a lemon. Their core asset is no longer their distribution network or their brand logo, but the verifiable integrity of their process. This suggests a radical new business model. Publishers should stop thinking of themselves primarily as destinations for readers, and start thinking of themselves as authenticated sources for AI. Instead of selling ads against articles, they should license their credibility. Imagine a system where an AI’s answer cites its sources, and payments flow to those publishers proportionally, like digital royalties. A few pioneering deals, such as the one between The Atlantic and OpenAI, point in this direction. The big AI labs have a strong incentive to play along: their products are far more valuable when grounded in reliable information.
Making this shift necessitates a revolution inside newsrooms, too. AI should be a tool for journalists rather than a replacement. It can be a research assistant, sorting through haystacks of data to find narrative needles. It can be a devil’s advocate, stress-testing arguments. But it must not be the writer. The moment readers suspect that the byline is a fiction, the trust that is the industry’s last redoubt evaporates. The line is a fine one. Management will crave the efficiency AI offers; staff will fear it makes them obsolete. Dealing with this conflict needs strong editorial rules not only profit motives.
Not every publisher will make it. The industry is splitting in two. On one side are trust-driven institutions that invest in reporting, fact-checking and judgment. Their strategy, like that of The Atlantic, is to hold their mission constant while changing everything else around it. On the other side are “content farms” that treat journalism as a low-cost commodity, chasing clicks with cheap, search-optimised articles. This model is supremely vulnerable to AI. Why pay a human to write eight basic stories a day when a machine can do it for pennies? The coming wave will wash these operations away.
This Darwinian process will be brutal, but it is necessary. The chase for viral scale, already a dubious strategy, will become a suicide note. The survivors will be those that understand their product is credibility rather than content. They will offer a service that the largest AI lab cannot authentically replicate: the application of principled human judgment to the messy world of facts. The meteor is coming. Publishers can curse it, or they can find a way to harness its fire. Their future depends on choosing wisely. ■
