From threat to revenue stream: Publishers test new models for the AI era

Judge Amit Mehta opened his remedies ruling in the Google search antitrust case with an observation: The rise of generative AI has already reshaped the industry.

“No existing rival has wrested market share from Google,” Mehta wrote. “And no new competitor has entered the market. But artificial intelligence technologies, particularly generative AI (‘GenAI’), may yet prove to be game changers.”

For publishers, AI feels less like a game changer and perhaps more like déjà vu. More than 20 years after Google built its business model on the back of their content — siphoning off billions in ad revenue — publishers now face another existential threat: AI search.

“It’s already happening,” says Josh Brandau, co-founder and CEO of Nota. Brandau spent nearly six years as the CMO and CRO of the Los Angeles Times. “We’ve seen every single time, if something’s easier but not as good, it will still win.”

With these dynamics emerging, tech players are delivering new innovations aimed at protecting publishers’ revenue by helping them leverage AI instead of being crushed by it.

“Publishers should treat AI not as a threat, but as a new monetization layer — one that complements existing revenue streams like subscriptions, advertising and syndication,” wrote Josh Peters, Bombora’s VP of global data partnerships and strategy, in a 125-page guide he created on setting an AI monetization foundation.

Peters — who spent over 11 years at The Washington Post and BuzzFeed — argues that publishers should leverage their content, data and infrastructure to generate income through licensing, automation and partnerships with AI platforms.

This comes as publishers are facing a new set of challenges because of AI search.

The traffic cliff

Human web traffic dropped 9% across the 2,000-plus sites tracked by TollBit in the second quarter of 2025, according to the company’s latest State of the Bots report. And bot traffic is getting a lot more common as large language models (LLMs) crawl the internet more aggressively.

One in every 50 site visits now come from a bot — up 400% from Q1, according to TollBit. Many bots disguise themselves as human users, mimicking actions like scrolling, clicking the page and even solving CAPTCHAs.

This bot traffic, along with AI chatbots creating zero-click searches, is massively bringing down referral traffic. In June 2025, Anthropic’s crawl-to-referral rate was 73,000 to 1, OpenAI’s was 1,700 to 1, and Google Search’s was 14 to 1, according to data from Cloudflare. A decade ago, Google’s rate was 2 to 1.

One publisher that signed up with TollBit was crawled by Google 16 million times in a month, receiving tens of millions of visitors in return. In that same 30-day period, AI companies scraped the site 13 million times and delivered just 650 visits.

“The AI companies claim to be just like Google — ‘Let us crawl you and we’ll send you traffic,’” TollBit CEO Toshit Panigrahi says. “This publisher thought otherwise after looking at the numbers. They figured, how many times does our site need to let you crawl us before we get 600,000 visitors? We will bankrupt by then just serving bots.”

The advertiser problem

This dynamic doesn’t just hurt publishers. Fake traffic wastes ad dollars and undermines trust.

“There will come a point where it will be impossible to tell apart bots and humans,” Panigrahi says. “That will be hugely detrimental to advertiser trust.”

Panigrahi adds that “this used to be an AI and publisher problem. This is about to become an AI, publisher and advertiser problem. And I don’t think the industry has realized this yet.”

TollBit is pushing a toll system for LLMs scraping and crawling the internet, essentially charging AI bots a “bot-scraping CPM” for access.

Cloudflare has made waves by blocking AI crawlers across 20% of the internet’s traffic. While Panigrahi says that strategy is helpful, it’s not enough. AI bots bypassed 13% of website block requests in Q2 of 2025, four times higher than Q1, according to the State of Bots report.

“If we only try to block the bots, then we just incentivize them just to pretend to be more human,” Panigrahi says.

Some AI firms are experimenting with revenue sharing. Perplexity, for example, will pay publishers per referral and inclusion in a search query within its subscription model. The program projects $42.5 million in annual payouts.

Bombora’s Peters says it’s a step in the right direction but not enough because “the pay scales are completely lopsided for publishers. That’s a drop in the bucket [compared to] their total valuation and what they’re paying publishers.”

The road ahead

In his search remedies decision, Mehta wrote that the emergence of GenAI led him to “gaze into a crystal ball and look to the future. Not exactly a judge’s forte.”

Panigrahi, on the other hand, is less equivocal.

“In five years, we’ll look back at this time and think it was crazy that we ever debated if bots should be paying for the content that they’re accessing online.”