No playbook, just pressure: Publishers eye the rise of agentic browsers

Tools like Perplexity’s Comet, OpenAI’s Atlas, The Browser Company’s Dia/Arc experiments, and Google’s Gemini-in-Chrome features hint at where this is going in 2026: browsing that feels more like an assistant running errands than a user clicking links. And that is exactly the kind of shift that can quietly collapse referral traffic and make publisher content the raw material rather than the destination. 

“AI is an accelerator for the decline of browser-based web experiences,” says Amir Malik, managing director at global professional services firm Alvarez and Marsal, and former publisher. “We’ve not even scratched the surface of AI summaries. We’re looking at a fundamental change in consumer behavior…A [publisher] destination site from a search to [a] landing page in the consumer journey is fundamentally being replaced with a prompt-based future,” he says.

For now, those fears remain largely theoretical. Agentic browsers (bar Chrome) are still nascent, lack mainstream adoption, and don’t yet have the scale needed to materially disrupt publisher traffic. One publisher exec, who agreed to speak on condition of anonymity, described the threat as a “3 percent problem” – too low a percentage to be in the top threats currently. “We have enough 10 percent problems to be worrying about,” as they put it. 

But after a year of watching AI Overviews reshape search behavior almost overnight, publishers say they’re paying close attention — wary of another shift that could quietly redraw how audiences find and consume journalism.

So far, there is no publisher playbook. Many have tested out the browsers to get a feel for how they work. Some publishers foresee an acceleration in the decline of link-based search traffic, but they also see potential in the agentic browsers’ abilities to help people work faster by getting them the information they need. 

Chrome’s agentic browser: the ‘only timeline that matters’

For the bulk of publishers, Google is, as ever, the one to watch. It’s already got agentic features within its Chrome browser, but that’s the tip of the iceberg, some say. “Google’s agentic browser – that’s the only timeline that matters,” as one publisher put it, who exchanged candor for anonymity. 

Currently, Chrome integrates several AI features but full agentic capabilities, where the browser autonomously performs complex tasks on a user’s behalf, are still in development and rolling out gradually. 

The same publishing exec predicted that once it nails the right utility for consumers in that experience, publishers will need to tune in fast. “Once you [consumers] can ask Chrome to do things for you, we all have to figure out what our websites look like when it’s an agent, not a human, browsing our websites through Chrome,” they said. 

The other newly launched AI browsers: Comet, Atlas, Dia, can already browse the web on a user’s behalf. And agentic visitors generally are causing some publishers trouble, with advertisers pulling spend because the tech isn’t there to separate agentic and human traffic. But it’s Google’s expected development in this area next year, that’s “alarming,” said the same exec. 

“We’ve spent a lot of effort having a direct connection with our consumers. But if the browser is disintermediating us – that’s a whole new level of disintermediation.”

Publishers want Google to draw a hard line between human traffic and agentic visitors — and to signal it clearly — so ads, measurement and experiences aren’t wasted on machines instead of people. “Google has a large advertising business, and I think some of their customers might care a lot whether those ads are being shown to AI or humans,” said a publishing exec. 

Agentic browsers as data ‘Trojan horse’ for LLMs

That strikes a chord with Anthony Katsur, CEO of the IAB Tech Lab. As AI agents begin browsing the web, he argues publishers will need ad systems that can tell the difference from a human pageview and adapt monetization logic accordingly. The idea would be to recognize AI-driven visits and avoid running standard ad calls, which serve no purpose for LLMs. And instead replace them with smarter, server-side decisioning designed for non-human traffic. 

Some of the new AI/agentic browsers have built-in ad and tracker blocking switched on – like  Perplexity’s Comet for example. That’s a concern for some publishers, although there are no signs to say others will follow suit. 

Privacy controls within Apple’s browser Safari have blocked measurement and ad calls, sometimes to the point where ads fail to function properly, noted Katsur. That’s partly why this year more major publishers will move their ad stack from the browser to server-side, via the IAB Tech Lab’s Trusted Server framework. In doing so, they’ll preserve more addressability, keep third-party tools working and reduce their exposure to browser-level blocking, he added. There are at least half a dozen publishers in the U.S., U.K. and other European countries like Germany, waiting to pilot Trusted Server, per Katsur, though he wouldn’t name them yet. 

No set playbook doesn’t mean total inaction, stresses Katsur. Although OpenAI and Perplexity’s browsers don’t yet have scale (Chrome dominates market share globally at approximately 70 percent), publishers should get under the hood of what exact data these browsers are retaining and what information they’re sending back to their large language models. “I would proceed with caution if you’re a publisher – I wouldn’t let them [Atlas and Comet] access my content, because I suspect that it’s a Trojan horse to crawl you through the browser,” he said. 

He warns that even at a small scale, they may cause problems for publishers if they’re ignored. “Publishers may not notice in the beginning, but the best way to stop things like this is early, don’t let it get traction,” he added.