News publishers are facing a paradox in the age of AI. While referral traffic from chatbots like ChatGPT is surging, this growth is dwarfed by the catastrophic collapse in clicks from AI-powered search summaries, according to a new report from web analytics firm Similarweb.
The July 2025 analysis found that while ChatGPT referrals grew 25-fold, overall organic search traffic to news sites plummeted by over half a billion visits from its peak. This has ignited a fierce debate over the value of this new, smaller stream of traffic.
The crisis is forcing a reckoning, prompting publishers to pursue lawsuits, deploy bot-blocking tools, and explore new payment systems to counter what many see as an existential threat to the web’s ad-supported model.
The Great Traffic Collapse: A Story in Numbers
The scale of the traffic decline is staggering. The Similarweb study reveals that the percentage of “zero-click” news searches, where users get answers from AI summaries without visiting a publisher, has jumped from 56% to nearly 69% in the year since Google launched AI Overviews.
This shift was accelerated by Google’s global expansion of AI Overviews, which began in early 2025 and fundamentally altered the search landscape by prioritizing direct answers over the traditional blue serach links. This trend is gutting the referral model that has sustained online media for decades.
Cloudflare CEO Matthew Prince has called this an “existential threat,” revealing data that shows OpenAI’s crawl-to-visit ratio for publishers has worsened from 250:1 to a staggering 1,500:1 in just six months.
The situation with other AI models is even more dire. Prince noted that the ratio for Anthropic’s AI has cratered from 6,000:1 to an almost unbelievable 60,000:1, meaning AI is consuming vast amounts of content while returning almost no traffic.
Quality Over Quantity: The AI Referral Paradox
Amid the search traffic carnage, a small but growing stream of referrals from AI chatbots offers a sliver of hope. ChatGPT referrals to news publishers grew 25-fold, from under 1 million in early 2024 to over 25 million in the same period of 2025.
This has sparked a debate on whether this new traffic is of higher quality. Media analyst Simon Owens noted that these visitors seem more engaged, citing his own newsletter stats showing a 9% conversion rate from ChatGPT referrals versus just 2.5% from Google search.
However, these benefits are not evenly distributed. Publishers like Reuters and the NY Post have seen significant referral growth from ChatGPT. In contrast, The New York Times, which is actively suing OpenAI for copyright infringement, has seen far less of an increase.
The disparity suggests that AI companies may be selectively directing traffic, rewarding friendly publishers while penalizing those who challenge their data practices. This raises questions about the long-term viability of relying on chatbot referrals as a sustainable revenue source.
Publishers Fight Back: From Lawsuits to Firewalls
Meanwhile, the industry is not standing still and has taken on the fight. Publishers are launching a multi-front war combining legal, financial, and technical strategies. On the legal front, a wave of lawsuits from major US newspapers and The New York Times targets AI developers for copyright infringement.
This legal pressure is being matched by financial demands. In Germany, media rights group Corint Media is demanding approximately €1.3 billion annually from Google for using journalistic content. Its Co-CEO, Markus Runde, stated, “We consider our calculation to be conservative. The actual value that Google derives from journalistic content is likely to be even higher.”, signaling a new push for direct compensation.
The industry’s frustration was captured by Danielle Coffey, CEO of the News/Media Alliance, who said, “Links were the last redeeming quality of search that gave publishers traffic and revenue. Now Google just takes content by force and uses it with no return.”. This sentiment is driving the adoption of aggressive technical countermeasures.
Cloudflare has emerged as a key ally for publishers. In a defiant stance, CEO Matthew Prince once quipped, “And you’re telling me, I can’t stop some nerd with a C-corporation in Palo Alto?”. The company has backed this up with powerful tools.
Cloudflare recently launched “Pay Per Crawl,” a system that blocks AI crawlers by default and allows publishers to charge for access using the long-dormant HTTP 402 “Payment Required” status code. This follows its earlier deployment of “AI Labyrinth,” a tool designed to trap and confuse unauthorized bots.
However, the fight is complicated by findings that Google uses content for its AI Overviews even from publishers who have opted out of general AI training. This suggests that simply blocking bots may not be enough to stop the largest players.
As the battle rages, the future of online publishing hangs in the balance. The symbiotic relationship between search and content is broken, and as OpenAI CEO Sam Altman has acknowledged, “I do think there will be areas where some jobs go away… there is going to be real pain here in many cases.”. The industry is now in a race to build a new model before the old one disappears entirely.
