Allowing publishers to exclude their content from Google's AI Overviews without affecting their presence in traditional search results is a "huge engineering project," according to a senior Google executive who addressed editorial and product leaders at a publishing conference on February 11, 2026. The admission came the same week the Financial Times announced a content licensing deal with the company - a pairing of signals that together illustrate just how complicated, and contested, the relationship between Google and the news industry has become.

Sulina Connal, Google's managing director for news and books partnerships in Europe, made the comments at the FT Strategies News in the Digital Age conference in London. Speaking to publishers, she described the scale of what a granular opt-out mechanism would involve. "Because AI is integral to how search works, implementing the new controls is a complex engineering, huge engineering project, and we have been highly intentional in our approach," according to Connal. "The idea is to focus on simple, scalable tools that you can use to manage your content."

The remarks drew immediate scepticism. Paul Bannister, chief strategy officer at Raptive, posted on LinkedIn that Google had committed to hiring "tens of thousands of engineers" and spending "literally a trillion dollars on chips and power," while simultaneously claiming it could not build tools to let publishers signal content exclusions. The post, shared on February 17, 2026, attracted 27 reactions and prompted pointed comments from across the advertising and publishing community. Wicky Sri, a startup operator, noted the irony that "AI can write thousands of lines of code, but somehow letting publishers opt out of AI Overviews is impossible." Adam Gallagher added, simply: "Yeah, not buying this excuse. They obviously have the resources."

What the controls would actually do

The core challenge Connal described is not about building a mechanism to exclude content from Google Search entirely - that already exists. What publishers have been demanding for months is something more specific: the ability to block their content from appearing in AI Overviews and AI Mode while still having their pages indexed and ranked in conventional search results.

This distinction matters enormously for business models built around organic traffic. Publishers who currently attempt to block Google's AI features face an all-or-nothing choice - either they allow full participation, or they opt out of "rich experiences" entirely, which substantially reduces their visibility in standard search results. That forced trade-off is what the proposed granular controls would resolve.

Google confirmed it is "exploring updates" to its controls, and Connal told conference attendees the company had "heard the need for more controls" consistently from publishers. She acknowledged that making informed decisions requires data, and that engagement metrics are central to any meaningful system. "We also understand that to allow you to make informed decisions about search AI features you need more information, and measuring users' engagement is key to that. We hear you, and we hear you loud and clear," according to Connal.

But critics, including commenters on Search Engine Roundtable, questioned the priority given to the project. One commenter, Vic Daniels, wrote that "if it is a huge engineering project then maybe Google should wind it back and stop using unfair/rip off AIOs until they are in a position to do the right/ethical thing." Another, Quetzal, said it "sure didn't seem like it was too much of a problem for them to set up the system to steal all that content."

Regulatory pressure behind the announcement

The timing of Connal's comments is not incidental. The UK's Competition and Markets Authority had already moved, on January 28, 2026, to propose binding conduct requirements under the Digital Markets, Competition and Consumers Act 2024, including a requirement that Google provide publishers with the ability to opt out of AI Overviews without losing search visibility. That proposal followed the CMA's designation of Google with Strategic Market Status on September 30, 2025, after a nine-month investigation that concluded the company holds substantial and entrenched market power in general search services.

Google had already responded on January 28 by announcing it was exploring opt-out controls, the same day the CMA published its consultation. Google's Product Management Principal, Ron Eden, stated the goal was to "protect the helpfulness of Search for people who want information quickly, while also giving websites the right tools to manage their content." But the conference exchange two weeks later made clearer what the company itself considers the difficulty: ensuring that any new controls affect only the AI search features, not the traditional indexing and ranking systems that underpin core search.

The CMA's consultation, open until February 25, 2026, also covers fair ranking demonstrations, choice screens on Android and Chrome, and data portability mechanisms. The publisher control requirement is classified as a Category 1 measure, slated for implementation in the first half of 2026, with more complex Category 3 interventions following from 2027.

A separate European Commission investigation, launched on December 9, 2025, is examining whether Google violated EU competition rules by using publisher content for AI purposes without appropriate compensation or viable opt-out mechanisms. Publishers in the EU have also brought their own formal antitrust complaints. Independent publishers filed with both the European Commission and the UK regulator on June 30, 2025, alleging that traffic declines of 34.5% or more followed the appearance of AI Overviews in search results.

The Financial Times signs on

Against this contested backdrop, Financial Times chief executive Jon Slade announced at the same London conference that the FT had signed a content licensing deal with Google covering a "series of AI pilot projects." The announcement was made on February 11, 2026.

"I really commend Google for showing up to the debate and being prepared to listen and act, and I'm encouraged by emerging licensing deals, by discussions of marketplaces grounded in high quality journalism and by growing recognition that original reporting and human judgement have value in an AI era," according to Slade. He described the arrangement as a "pioneer" move, but was careful to frame expectations. "We don't yet know precisely where this will take us. It is a pilot, after all," according to Slade.

The FT joins a group of publishers that already signed similar deals, including The Guardian and The Washington Post. The deals, first announced in December, are believed to include cash payments alongside "extended display rights and content delivery methods like APIs." Connal said Google is "testing different deals and different approaches to deals" and intends to continue, promising "more focused deals that help advance the user experience."

The deal structure has been described as selective - large publishers with legal and commercial resources to negotiate individual arrangements gain access to payments and experimental features, while smaller outlets face the same traffic pressures without comparable financial arrangements. Google launched AI-powered article summaries for a select group of publishers in December 2025, including Der Spiegel, El País, Folha de S. Paulo, Infobae, Kompas, The Times of India, The Washington Examiner, and The Washington Post. The FT was not among the initial group.

Slade's language at the conference was notably pointed about what this moment represents. "It was, I think it's fair to say, a long road," according to Slade. He described the opportunity as a chance to "move from a system that rewards facsimiles and clickbait journalism to ensuring that value flows to the news organisations that check and recheck stories before publication."

Traffic data and the preferred sources feature

Connal also addressed traffic directly, citing a feature Google introduced in the United States last summer: Preferred Sources, which allows users to select which news outlets appear more prominently in the Top Stories box when those outlets publish "fresh and relevant" content. The feature expanded globally to all English-language users in December 2025.

Publishers have already begun asking their own readers to add them as preferred sources. The Guardian, for example, told users directly: "By selecting the Guardian as a preferred source you'll have more control over what shows up in your search results without having to rely on the algorithm alone."

Connal also outlined a forthcoming Gemini app feature that will "prioritize and highlight links to publishers with which a user has a subscription," with rollout expected later to AI Overviews and AI Mode. This feature, described in an X post by SEO analyst Glenn Gabe on February 16, 2026, positions subscription relationships as a signal within AI surfaces - a meaningful development for publishers with paying reader bases.

Meanwhile, Arun Venkataraman, who leads Google's global industry research partnerships with the news ecosystem, told a separate event in London hosted by Beehiiv on Thursday, February 13, that Google's internal data does not show a decrease in total traffic referrals from search. "Overall traffic referrals from search have actually not decreased based off of our internal data. But what is happening is the sources that that traffic referral is going to are more disparate," according to Venkataraman. He suggested that some traffic that previously went to large publishers may now be going "to emerging news sources that might be more relevant in certain contexts."

This interpretation differs substantially from Chartbeat data published by Press Gazette in January, which showed global traffic to publishers from Google Search dropped by approximately a third in 2025. Venkataraman acknowledged these studies track a limited number of publishers or specific geographies rather than the full ecosystem.

What this means for the marketing community

For digital marketers and advertising technology professionals, the implications are layered. The absence of a workable AI Overviews opt-out creates measurement problems. Without the ability to test content in and out of AI features independently, publishers cannot generate clean data on whether AI Overviews help or harm their business. That uncertainty cascades into advertising inventory planning, programmatic floor prices, and publisher-side yield management.

Research from Seer Interactive, published November 4, 2025, found organic click-through rates for informational queries featuring AI Overviews fell 61% since mid-2024, while paid click-through rates on those same queries dropped 68%. The study covered 3,119 informational queries across 42 organizations, including 25.1 million organic impressions and 1.1 million paid impressions between June 2024 and September 2025.

The Penske Media antitrust case, with its 56-page opposition to Google's motion to dismiss filed February 12, 2026, cited projections of advertising revenue loss and traffic decline ranging from 20% to 60%. Bain & Company research cited in the filing concluded that 60% of Google searches now end without a click. Similarweb reported an 83% zero-click rate for searches where AI Overviews appear.

These figures frame the engineering timeline as a business-critical issue. Each month the opt-out mechanism remains under development is another month publishers cannot make data-backed decisions about content strategy. Connal's statement that Google has been "highly intentional" in its approach is unlikely to satisfy publishers facing immediate revenue pressure.

The two-tier system and its cost for smaller publishers

While Google's negotiations with the Financial Times and other major news organisations attract most of the conference coverage, a quieter and more consequential story has been unfolding for the publishers who were never invited to the table. The commercial pilot Google launched in December 2025 - paying select outlets for AI content access and experimenting with AI-generated article overviews on their Google News pages - was extended to Der Spiegel, El País, The Guardian, The Washington Post, and a handful of other globally recognised titles. The partnership structure creates a two-tier system where major publishers receive compensation and experimental features while smaller outlets face the same traffic declines without any financial buffer.

For independent publishers, this is not a theoretical concern. The numbers have been accumulating for two years. Google Web Search has fallen from sending 51% of traffic to news publishers in 2023 to just 27% in the fourth quarter of 2025, according to an analysis of more than 400 publishers worldwide by NewzDash, published December 23, 2025. Google Discover, the algorithmically curated smartphone feed, now accounts for 67.51% of Google's traffic to news organisations - a source publishers have far less ability to optimise for, and one that can disappear overnight.

That last point is not hypothetical. Google's December 2025 core update, announced December 11, triggered Discover traffic collapses within 48 hours for hundreds of publishers. Some reported 70 to 85% declines in daily visitor counts during what would normally be their most commercially valuable period of the year. One operator quoted at Search Engine Roundtable described Discover traffic dropping 98% in the days before the update was even officially announced. These sites had no recourse, no account manager, no licensing negotiation in progress. They simply absorbed the loss.

The click-through rate data makes the structural shift even clearer. Ahrefs published research on February 4, 2026, showing AI Overviews now correlate with a 58% reduction in click-through rates for top-ranking pages - nearly double the 34.5% decline the same company documented in April 2025. For informational keywords specifically, average position-one click-through rates fell from 0.056 in March 2024 to 0.031 in March 2025. Keywords that triggered AI Overviews dropped even further, from 0.073 to 0.026 over the same period. The content still ranks. Users still see it summarised. They just no longer visit.

The advertising implications are direct. Publishers generate revenue when users load pages, view display ads, or click affiliate links. Impressions registered by Google on its own interface - where an AI Overview extracts and presents the key information from a review or guide - generate nothing for the site that produced it. Gisele Navarro, managing editor of HouseFresh, an independent site that independently purchases and tests air purifiers before reviewing them, demonstrated this precisely when she shared internal analytics showing her content's visibility within AI Overviews had increased while actual site visits had declined. Her observation was blunt: "Sites don't generate revenue unless users click through and see ads or buy products."

HouseFresh's experience is also instructive in another way. After the September 2023 Helpful Content Update, the site's Google traffic fell 95%, from 4,000 daily visitors to approximately 200. Two years of sustained effort - improving content, building a YouTube channel, and a collaboration with Linus Tech Tips that expanded brand recognition - eventually resulted in a partial recovery visible by October 2025. But the operators themselves described the outcome as "hard work and a lot of luck." The recovery was exceptional. Most sites hit by the 2023 update did not recover at all.

Geekflare, a technology publisher founded in 2015 that had built to 6 million monthly pageviews, reduced its workforce from 53 employees to 2 people following a 90% traffic decline. Spanish automotive website Test Coches lost approximately 3 million monthly readers. Dave Bouskill and Debra Corbeil, who launched travel blog The Planet D in 2008, watched traffic fall 90% after AI Overviews began reproducing travel tips they had developed over years on the road - including, as they noted bitterly, their own Canadian slang. "I do feel betrayed by Google," Bouskill said. They stopped updating the blog entirely.

Nick Fox, Google's SVP of Knowledge and Information, was asked directly about these situations during a December 15, 2025 podcast interview, five days after Google announced its selective commercial partnerships. His answer - that optimising for AI search is "very similar, I would say, the same as how to perform well in traditional search" and that it "really does come down to build a great site, build great content" - drew sharp criticism from industry observers. Fox acknowledged that "for sure there are sites that are struggling" and said he had "empathy for those sites," but offered no structural proposals beyond the existing traffic-and-links model. He explicitly declined to endorse standardised licensing frameworks that would give smaller publishers a path to negotiate compensation.

This is the gap that defines the current situation for independent publishers. Large organisations with legal teams and commercial relationships can negotiate bespoke arrangements with Google. The Financial Times took years to reach its pilot deal, and even that is described as experimental. Smaller outlets cannot replicate that process. They face the same algorithmic systems, the same AI-generated summaries consuming their content, and the same declining click-through rates - but without any mechanism for compensation or meaningful control.

A survey of 280 news executives published January 17, 2026, found that global Google Search traffic to publishers had declined 38% in the United States and 33% globally between November 2024 and November 2025. UK website traffic growth collapsed 86% since Google's AI search rollout, according to Tank research tracking 800 companies across 16 sectors. Average monthly organic traffic growth across analysed industries dropped to 3.7% from 26.3% the previous year - a fall of 22.6 percentage points. The survey respondents expected a further 43% reduction in the coming year.

The opt-out mechanism that Sulina Connal described as a "huge engineering project" would, if built correctly, give all publishers - not just those with negotiating leverage - a meaningful choice about how their content is used. Publishers have been requesting this specific capability since at least October 2025, when the absence of granular AI feature controls became a public flashpoint. The ability to test content in and out of AI Overviews independently would, at minimum, generate the kind of data publishers need to make informed decisions about their content strategies. Without it, they are navigating blind, watching traffic figures move without being able to determine which Google surface is responsible for which change.

What the conference in London illustrated is that Google has two parallel conversations underway. One is with major publishers, conducted through licensing negotiations, paid pilots, and conference appearances - a commercial dialogue about the future of AI-mediated content distribution. The other is with the broader publishing ecosystem, conducted through regulatory consultations, public statements about engineering complexity, and quarterly algorithm updates that restructure traffic patterns with little notice. These two conversations have very different participants, very different stakes, and so far, very different outcomes.

Timeline

Summary

Who: Sulina Connal, Google's managing director for news and books partnerships in EMEA, and Jon Slade, CEO of the Financial Times, speaking at the FT Strategies News in the Digital Age conference in London. Arun Venkataraman, Google's head of global industry research partnerships with the news ecosystem, addressed a separate Beehiiv-hosted event the same week.

What: Connal described building an AI Overviews opt-out mechanism that does not affect traditional search visibility as a "huge engineering project," while Slade announced the FT had signed a content licensing deal with Google covering a series of AI pilot projects. Connal also outlined upcoming features including a Gemini app tool that will highlight links to publishers with which users hold subscriptions, later expected to extend to AI Overviews and AI Mode.

When: Connal and Slade spoke on February 11, 2026, at the FT Strategies News in the Digital Age conference. Venkataraman spoke on February 13, 2026. Reporting on the remarks circulated from February 16 to February 17, 2026.

Where: The FT Strategies conference was held in London. Venkataraman's comments were made at a separate London event hosted by Beehiiv. Regulatory activity referenced in the piece originates from UK and EU authorities.

Why: Publishers have been demanding the ability to exclude their content from AI Overviews and AI Mode while retaining traditional search visibility, citing significant traffic and revenue declines associated with AI-generated summaries. The CMA proposed binding conduct requirements on January 28, 2026, creating regulatory pressure on Google to act. The FT's licensing deal illustrates the parallel commercial track Google is pursuing with major publishers, even as the technical infrastructure for broader publisher controls remains unresolved.

Share this article
The link has been copied!