The AI Threat to Journalism

AI is posing an existential threat to journalism by siphoning off traffic and revenue, with users clicking through to source articles at half the rate when AI summaries appear. News organizations are grappling with how to adapt, seeking fair compensation and new strategies to survive in an increasingly AI-driven information landscape.

The storm has been gathering for years, a relentless erosion of the foundations supporting independent journalism.

Now, a new tempest is upon us, one powered by algorithms and the voracious appetite of generative artificial intelligence, threatening to sweep away what remains of an already battered media ecosystem.

The very act of seeking information online, once a reliable conduit for news organizations to connect with readers and advertisers, is undergoing a profound transformation, and the implications are nothing short of existential.

For decades, the internet promised boundless reach, a digital town square where news could thrive.

Instead, it delivered a landscape dominated by tech giants, siphoning off advertising revenue and forcing publishers into a precarious dance with search engine optimization.

Just as newsrooms began to recalibrate, pivoting towards direct reader support through subscriptions, a new, more insidious challenge emerged: AI assistants like ChatGPT.

These sophisticated tools, designed to provide instant answers and summaries, are effectively cutting off the oxygen supply to news websites, depriving them of the traffic that fuels both ad revenue and, crucially, subscription conversions.

The numbers, though still emerging, paint a grim picture.

A recent Pew Research Center study reveals a chilling trend: when AI-generated summaries appear in Google searches, users click through to source articles at half the rate compared to traditional searches.

Half.

Imagine the immediate, devastating loss of potential readers, the advertising impressions vanishing into thin air.

“The next three or four years will be incredibly challenging for publishers everywhere. No one is immune from the AI summaries storm gathering on the horizon,” warned Matt Karolian, vice president of research and development at Boston Globe Media.

His stark advice — “Publishers need to build their own shelters or risk being swept away” — underscores the urgency of the moment.

This isn’t merely a shift; it’s an acceleration towards “an entirely different web,” as Northeastern University professor John Wihbey aptly puts it.

The traditional model, already strained by the dominance of Google and Meta, relied on traffic to convert casual visitors into loyal subscribers.

But if AI intercepts that initial engagement, the path to subscription becomes far more circuitous, if not entirely blocked.

While Karolian noted a modest trickle of new subscribers signing up through ChatGPT for The Boston Globe, he quickly tempered expectations: these numbers are “incredibly modest compared to other platforms, including even smaller search engines.” Other AI tools, like Perplexity, yield even fewer results, suggesting that the AI pipeline for reader acquisition is currently more a drip than a gush.

In the face of this tsunami, news organizations are scrambling for new survival strategies.

The old adage of Search Engine Optimization (SEO) is giving way to Generative Engine Optimization (GEO).

This involves meticulously structuring content for AI models — clear labeling, logical organization, comprehensible text, and a robust presence on social networks and forums like Reddit, which AI companies diligently crawl.

It’s a pragmatic adaptation, a recognition that if you can’t beat them, you must at least try to be understood by them.

Yet, a fundamental moral and economic quandary persists: “Should you allow OpenAI crawlers to basically crawl your website and your content?” asks Thomas Peham, CEO of optimization startup OtterlyAI.

Having been burned by the aggressive data harvesting practices of tech giants in the past, many news publishers initially chose to block AI crawlers, asserting their intellectual property rights.

The rallying cry is clear: “We just need to ensure that companies using our content are paying fair market value,” declared Danielle Coffey, head of the News/Media Alliance.

This stance has led to some breakthroughs, with licensing agreements emerging between major players like The New York Times and Amazon, Google and the Associated Press, and Mistral and Agence France-Presse.

These are vital steps towards recognizing the immense value of original reporting.

However, the path to a fair settlement is fraught with legal battles, most notably The New York Times’ blockbuster lawsuit against OpenAI and Microsoft.

Publishers find themselves on the horns of a dilemma: blocking crawlers protects their content but severely limits their potential exposure to new readers in an increasingly AI-driven search landscape.

Peham observes that “media leaders are increasingly choosing to reopen access,” a testament to the desperate need for visibility, even if it means risking further content appropriation without adequate compensation.

Even with open access, success is far from guaranteed.

OtterlyAI data indicates that media outlets account for a mere 29 percent of citations offered by ChatGPT, lagging behind corporate websites at 36 percent.

This stark disparity highlights another critical concern: while Google search traditionally prioritized authoritative and reliable sources, “we don’t see this with ChatGPT,” Peham notes.

This has profound implications for the quality and trustworthiness of information.

The Reuters Institute’s 2025 Digital News Report indicates that roughly 15 percent of people under 25 now use generative AI to get their news.

This trend, coupled with AI’s often-opaque sourcing and occasional factual inaccuracies, risks creating a generation confused about information origins and credibility — a troubling echo of the early social media era.

Ultimately, the core issue remains unchanged: “At some point, someone has to do the reporting,” as Karolian succinctly put it.

Without the tireless work of journalists — investigating, interviewing, fact-checking, and crafting original narratives — these sophisticated AI platforms would have nothing of substance to summarize, nothing to learn from, nothing to regurgitate.

They are, in essence, parasitic, drawing their lifeblood from the very content creators they threaten to undermine.

There is a glimmer of hope.

Google, perhaps recognizing its own long-term dependence on quality information, is reportedly developing partnerships with news organizations to feed its generative AI features.

“I think the platforms will realize how much they need the press,” Wihbey predicted, a sentiment that offers a fragile optimism.

But whether that realization dawns soon enough, and whether it translates into equitable partnerships that truly sustain struggling newsrooms, remains the most pressing, unanswered question in journalism today.

The storm is here, and the resilience of the news industry hangs in the balance.