AI’s Impact on Journalism: Winners, Losers, and the Vanishing Middle 

Artificial Intelligence is reshaping the news industry, but not in the ways most headlines suggest. The real story is more complex and more revealing, argues Ezra Eeman, Strategy and Innovation Director at Dutch public broadcaster NPO and leader of WAN-IFRA’s Global AI and Media initiative. 

This article distills key insights from Twipe’s first AI Frontrunners in News podcast, where Eeman outlined four underlying dynamics that cut through the noise and show what AI really means for the future of news. 

The Dream of Getting Paid by AI Is Built on Broken Models 

For many publishers, the hope is simple: that AI companies will one day pay for the journalism fueling their models. But as Ezra Eeman points out, the most-discussed proposals don’t hold up when tested against scale or reality. 

Pay-per-Crawl 
In this model, publishers charge a micro-fee each time an AI system ingests an article. On paper, it sounds straightforward; in practice, it collapses under volume, particularly when viewed from the perspective of AI companies.  

As Eeman notes, “if you know that ChatGPT is processing one billion queries a day, what’s the price point for a crawl? Even at a micro fee, it would mean billions per year that would need to be paid out.” 

In practice, if AI models were forced to pay-per-crawl, they would either scrap free access, raise subscription fees dramatically, or simply find alternative ways to get the content for free. 

Attribution Splits 
Here, revenue is divided based on which sources contribute to an AI-generated answer. Companies like ProRata have experimented with this model, but its flaws are clear. 

First, attribution remains a niche approach no major AI player has adopted, creating a “supply and demand” problem. Second, the logistics are unworkable: when an output draws on hundreds of sources, expecting one platform to assign and distribute microscopic revenue shares fairly is nearly impossible. 

Do you trust one party to define the attribution mechanism, or will there be different attribution mechanisms depending on the platform? Would you trust Google to determine which part you get from the pie?

Ezra Eeman

Strategy and Innovation Director, NPO

The Spotify Model 
Inspired by music licensing, this approach assumes publishers could earn recurring payouts as content is “played.” Perplexity, for instance, floated the idea with its Comet Plus subscription model. But news isn’t music. As Eeman notes, “songs are played many times over time, while an article is read once.” At scale, the lifetime value of a story would be so low that the revenue becomes meaningless. 

Eeman’s conclusion is blunt: while publishers may hope for compensation, these models won’t deliver meaningful revenue. The business case for AI in news lies elsewhere. 

The News Industry Won’t Die, But the Middle Ground Will Vanish 

Despite unfinished business models, Eeman’s vision for the future of news is ultimately optimistic, just not for everyone. He predicts not an industry-wide collapse but a “hollowing out” of the middle. The news landscape will restructure into three distinct tiers: 

• Who Survives: The big, trusted brands will endure. Their authority and brand recognition will ensure they remain direct destinations for users seeking reliable information. 

• Who Thrives: A vibrant new ecosystem of “smart, smaller operations” will flourish. This includes niche publications and “nano businesses”—like individual creators or small newsrooms that leverage AI—that serve a specific, dedicated audience with hyper-focused content. 

• Who Struggles: “Everything in between” will face the biggest challenge. Medium-sized news organizations, often reliant on now-failing advertising models and lacking the scale of the giants or the agility of the niche players, will be squeezed out. 

This dynamic highlights the importance of the enduring power of brand trust in a crowded market. As Eeman explains with an analogy, “Why do I go to the local bakery? Because I know they just have better croissants than the supermarket, which has everything.” The big brands are trusted supermarkets; the niche players are trusted bakeries. News organizations must choose to be either a large, trusted destination or a small, specialized provider, because attempting to occupy the “medium will not be worthwhile.” 

Forget Revenue—AI’s Real Return on Investment is Hiding Elsewhere 

Recent headlines, fueled by a widely circulated MIT report, created a media narrative that up to 95% of enterprise AI experiments fail. For Eeman, however, argues this is a classic case of the industry latching onto a sensational headline without proper scrutiny. He is deeply skeptical of the report’s methodology, noting it was “quite light survey” based on interviews with just 53 people.  

A more grounded picture emerges from Eeman’s own WAN-IFRA survey, which offers a crucial distinction. His research found that while only 9% of AI use cases delivered a clear return on investment (ROI) in terms of direct revenue, the results were far more positive when measuring for productivity and efficiency.  

For newsrooms, AI is already proving its value in streamlining workflows, automating repetitive tasks, and freeing up journalists for higher-value work. The true ROI of AI isn’t in a new revenue stream (at least not yet). Rather, it’s in the day-to-day operational efficiencies that headlines often ignore. 

In a World of Bots, True Human Journalism Becomes a Premium 

Eeman argues that, in a world saturated with AI-generated content, the core functions of human journalism become more distinct and valuable than ever.  

He identifies several core functions that AI cannot replicate, which will become premium differentiators for news organizations: 

• Holding Power to Account: AI systems are pattern-matchers, not investigators. They are not designed to challenge authority or conduct the difficult work of accountability journalism. 

• Human Judgment and Intent: A unique tone of voice, a distinct personality, and the critical editorial choices behind why a story is being told are uniquely human traits that build connections with an audience. 

• “Boots on the Ground”: There is no substitute for live, firsthand reporting. An AI can describe a forest fire based on data, but it cannot replace a journalist on the ground capturing the human reality of the event. 

• Building Trust: In a noisy, AI-filled information ecosystem, the acts of verification, fact-checking, and building a trusted brand become the most critical assets a news organization can possess. 

Who’s holding power to account? I don’t think AI systems will do that.

Ezra Eeman

Strategy and Innovation Director, NPO

The Real Questions to Ask 

AI will not kill journalism, but it will reshape it. As Ezra Eeman’s analysis shows, publishers cannot pin their hopes on flawed compensation models or fear being replaced by machines. Instead, the industry’s future will be defined by two forces: the operational efficiencies AI already delivers and the enduring value of human-driven journalism. 

This leads to the ultimate strategic question every news leader, journalist, and reader must consider: As AI generates an infinite sea of content, what unique human value will you consider worth paying for?