.jpg)
As digital publishers in India and around the world face declining traffic and shrinking ad revenues — a trend e4m has tracked since the launch of ChatGPT — a new research paper - The News Pipeline - authored by Stanford graduate researchers Alex Webb and Amrutha Nair, provides one of the most compelling analyses yet of what’s truly at stake.
Their sobering conclusion: the integration of AI into news distribution has the potential to fundamentally disrupt and destabilize journalism’s core business models.
Webb and Nair open with an unsettling hypothesis — what if content licensing deals signed with AI platforms aren’t a new lifeline, but instead a slow-fade into irrelevance?
“News organizations and operators of large language models have signed scores of content licensing deals over the past two years, mostly flat fees in return for access to all their content. For publishers, it looks like incremental revenue. But such deals might also hurt the news business. Because if artificial intelligence agents become the main way that consumers get their news in three to five years’ time, then they won’t visit the publishers’ own platforms.” they write.
The authors further note, “Publishers don’t just want revenue. They want traffic, and the opportunity to build and retain relationships with their readers.” But in the emerging AI-mediated ecosystem, even that modest ambition appears threatened.
“When AI summarizes the news, 99.7% of readers never visit the original sites. This risks being more than just another disruption, but an existential reckoning for journalism. AI might completely reshape both distribution and the fundamental rules of value,” Webb stated in a Linkedin post, sharing the study.
Notably, Nair had served as the head of Entertainment and Ad Sales in Disney Star in India, and Webb had been a TV commentator, Editorial Lead and Radio Anchor at Bloomberg in the UK till last July.
Drawing on extensive anonymised interviews with publishers, platform representatives, and tech startups, the paper dissects the evolving interplay between content creators and AI aggregators.
Their findings resonate sharply with ongoing developments reported in exchange4Media throughout the past year, particularly the steady erosion of loss of web traffic to news websites due to Generative AI, Google algorithm changes, SGE rollouts, and a dramatic fall in web traffic among major Indian and global news sites.
Reuters-Yahoo Deal in 1995
Researchers point to the Reuters’ 1995 deal with Yahoo—a pivotal moment that made news free at the point of consumption. In exchange for publishing its feed on Yahoo, Reuters received ad revenue and a $1 million stake in the platform. While the move suited Reuters’ licensing-driven model, it triggered a broader shift that pressured newspapers to drop paywalls, undermining subscriptions and eroding direct reader relationships.
“Reuters did not have the same business model (sales of physical newspapers and advertising) as newspapers, for whom the same decision carried enormous opportunity costs: it undermined their print sales and subscriptions, devaluing their core product and threatening their primary revenue streams,” the study states.
“The moment one major player offered content for free, it signalled to audiences that this was the market rate. From an economics standpoint this is a textbook Bertrand-competition setting: online news has very high fixed costs of creation but near-zero marginal cost of distribution. Price therefore was bid down to marginal cost (≈ $0), even though average cost is far above zero, so firms cannot recoup their fixed investment. The result is a classic market failure,” authors analysed.
The report warns of a similar risk today. “Once one outlet defects, the dominant strategy for the rest is to match the free model themselves.” The AI era risks repeating that historical mistake—only now, the disruption is faster, algorithmically scaled, and harder to reverse.
Impact of social media
Around the early 2010s, readers began relying on third-party platforms like Facebook for news discovery—often stumbling upon headlines while browsing for something else. As social media became the dominant gateway, news organizations lost control over distribution. By 2023, only 22% of online news consumers accessed news directly via websites or apps, compared to 25% through search and 29% via social media , the study noted, citing a previous research.
This shift introduced three layers of gatekeeping: newsmakers, platforms, and users—who determined what got shared or seen. As platforms imposed uniform formats for a seamless experience, news content was often bundled with user-generated posts and dubious sources. Readers rarely clicked through, reducing stories to snippets or clickbait. The result: commoditization of content, weakened ad and subscription models, and a slow erosion of news brands' authority and credibility, the paper noted.
RAG: A real-time threat
The authors focus on retrieval-augmented generation (RAG), where AI retrieves real-time external information before generating responses. Unlike models trained on static data, RAG accesses fresh web content, which could benefit publishers with timely reporting. However, AI’s clean, self-contained answers reduce user visits to original sources, making RAG both an opportunity and a threat.
Publishers relying on subscriptions face high risks. “Search and social gave us traffic; AI doesn’t,” a media executive noted. Without traffic, conversion funnels collapse.
The disruption stems from power and business model asymmetry. AI companies treat publisher content as “supply chain input” with little return. Licensing fees are often flat or vague, yielding minimal referral traffic—0.74 clickthroughs per 1,000 AI interactions versus 8.6% from traditional search.
Experts view
Commending the study, Sandeep Amar, founder, pdlab.me, observes, "Current AI licensing deals offer appealing short-term revenue but mask an existential threat to subscription-based journalism. Without sustainable business models that preserve direct reader relationships, AI risks undermining the very incentive to produce quality news—creating a market failure where journalism's positive externalities for society are undervalued and ultimately under-provided."
Echoing this concern, a news publisher adds, “A long, tough road lies ahead as AI reshapes news. Disruption to revenue is inevitable and stability will take time. Survival depends on innovation, creativity, and experimentation. News aggregators will likely be hit first.”
.jpg)
