Google AI Overviews are overtaking news boxes in search

Google is not just layering AI on top of news results. In a growing share of U.S. search results, AI Overviews are taking the place of the standalone News Box as the first thing users see. According to Trisolute News Dashboard data covering October 2025 through August 2026, the share of SERPs showing only a News Box fell from 75% to 34%, while the share showing only an AI Overview rose from 5% to 33%. When both features appeared, the AI Overview ranked above the News Box 96% of the time on desktop and 87% on mobile. For publishers, SEO teams, and brands that depend on Google visibility, the practical question has changed: not just can you win the module, but can you shape the answer.

• Standalone News Box visibility dropped sharply across the U.S. dataset, while AI Overview-only visibility surged.
• When both formats appeared, AI Overviews almost always took the higher position, especially on desktop.
• The shift matters most on topics Google can summarize, not just on breaking news.
• Teams now need separate reporting for rankings, answer-layer presence, citations, and traffic impact.

What exactly changed in Google’s results?

The short answer is simple: a large part of Google’s news-adjacent visibility is moving from a familiar news module into a generated answer layer. The Trisolute News Dashboard review found two major changes over the October 2025 to August 2026 window. News Box-only SERPs lost ground fast, and AI Overview-only SERPs gained it almost as fast.

That does not mean Google news visibility disappeared. It means the format changed. The old Top Stories or “Also in the news” style box no longer has the same monopoly on prominent placement. In late 2025 and early 2026, the pattern was relatively stable. Then summer 2026 brought a clear expansion in AI Overview-only SERPs.

There is also an important nuance here. This is not a one-cause explanation for publisher pain. The same dataset review noted that many publishers were already dealing with organic declines before AI features expanded, and social platforms were also taking more space in search. That matters because it stops the analysis from turning into a lazy “AI killed traffic” story. Google’s answer layer is a major shift, but it is arriving inside a search ecosystem that was already getting tougher.

A concrete example makes the change easier to see. If a publisher used to compete for a Top Stories box on a query with clear news value, several outlets could appear in the same visible block. As AI Overviews expand, Google can replace that multi-source box with a summarized answer that sends attention first to its own synthesis, not to the list of publishers below.

Why are AI Overviews pushing past News Boxes?

Because Google appears to be giving the generated answer layer more editorial weight than the traditional news module. The strongest jumps in the dataset lined up with Google core update periods, specifically March 27 to April 8 and May 21 to June 2. That does not prove causation, but it does show the expansion happened during moments when Google was already recalibrating result layouts and ranking behavior.

The more important product signal is this: during summer 2026, Top Stories carousels started appearing inside AI Overviews. In practice, Google often seemed to choose between an AI Overview with a Top Stories carousel and a standalone Top Stories News Box, rather than showing both prominently as separate features. That either-or pattern is the real story. If the answer layer becomes the container, the News Box stops being the default gateway to visibility.

There is a second-order effect too. A generated answer gives Google tighter control over framing. A News Box presents a set of headlines and lets the user choose. An AI Overview summarizes first, then decides which supporting links deserve attention. That is a different distribution model. It also changes what “top of SERP” means, because being present somewhere below a generated answer is not the same as being the main object on screen.

One example from the dataset review is especially revealing on desktop. A traditional Top Stories News Box can show several publishers in a single view. By contrast, a Top Stories carousel inside an AI Overview may show only three slots before the user has to click forward. Even if your publication remains included, the inventory becomes tighter.

This is also why the shift matters beyond news. BotRank recently looked at how Google AI search now shows more links, which sounds positive on paper. But more links inside a generated answer do not reverse the basic power shift. The answer layer still decides what gets summarized, which links appear close to claims, and which sources are visually secondary.

Why do desktop and mobile still behave differently?

Desktop still gives AI Overviews a stronger ranking advantage, while mobile preserves a little more room for News Boxes. Across the study period, when both features appeared, AI Overviews ranked above News Boxes 96% of the time on desktop versus 87% on mobile. That is not a trivial gap. It tells us the same query can create a different visibility battle depending on device.

Mobile is not exactly safe territory for publishers, though. The review showed that AI Overviews on mobile peaked at a 91% above-News-Box rate in November 2025, then declined to 83% by August 2026. In other words, News Boxes held their ground better on mobile, but the broader direction still favored the AI layer. August also marked the widest device gap outside January, with desktop running 13.2% higher than mobile for AI Overview placement above News Boxes.

The format difference helps explain why. On desktop, Google has more room to render a large answer panel, related links, and a tighter news carousel inside the same generated unit. That makes the AI Overview feel like the center of the page. On mobile, limited space can still help modular news elements hold attention, at least for now.

A concrete example from the study is slot density. A desktop News Box can expose multiple publishers in one glance, which gives users several headline choices immediately. A desktop AI Overview with a smaller carousel cuts that exposure down. For publishers, the problem is not only whether they are present. It is how many chances users get to notice them before Google asks them to engage with the synthesized answer instead.

What does this mean for publishers and brand teams?

It means visibility is now split across at least four layers: whether an AI Overview appears, where it sits, which sources feed it, and whether it still drives visits. If you only track rankings or Top Stories wins, you can miss the more important change happening above them. This is where source citation becomes an operational metric, not a vanity one.

For publishers, the immediate risk is obvious. A query that used to surface a clean set of headlines may now surface a summary first. For brands, the lesson is broader. Any content strategy built around explainers, comparisons, educational pages, or lightly newsy evergreen topics now has to assume Google may summarize before it sends traffic.

That does not make all content investment irrational. The dataset review made a useful point: if content that is easy for an AI Overview to answer still supports newsletters, apps, or YouTube distribution, it may remain worth producing. That is an important nuance. Search click value is changing, but content value can still exist elsewhere.

The measurement problem is where many teams will struggle most. If an AI Overview cites you and sends visits, your analytics may still understate the impact. BotRank covered that in AI Overview traffic is real, but GA4 may be hiding it, which showed how answer-layer clicks can be misclassified. And if your team is still collapsing mentions, citations, rankings, and clicks into one score, the framework in AI visibility metrics: the signals most brands still miss is the right correction. These are related outcomes, but they are not the same event.

A simple example: a publisher might lose standalone News Box share on a topic, gain a citation inside the AI Overview, and still see weaker CTR because the answer satisfied more users on the SERP. That is not a contradiction. It is the new search math.

BotRank’s Take

This story looks like a publisher problem, but it is really a search measurement problem first. Once Google moves the answer layer above a legacy module, classic ranking reports stop telling the full truth. You need to know which prompts trigger the answer, whether your brand or publication is present, and which pages are actually shaping the response. That is why BotRank’s AI Visibility feature matters in this context. It lets teams run a stable set of prompts across AI search environments and compare how visibility changes over time. Then Source Analysis adds the missing layer: which pages and domains sit behind those answers, and whether the cited pages truly mention the brand or support the claim. For teams dealing with shrinking News Box exposure, that distinction is not academic. Presence in the answer is one thing. Control over the evidence is what makes visibility useful.

How should teams adapt their GEO playbook now?

They should treat this as a Generative Engine Optimization (GEO) problem, not just a news-module problem. The right response is not panic. It is better segmentation, cleaner reporting, and stronger answer-ready pages.

• Separate trigger rate from ranking. First measure how often AI Overviews appear for your priority topics. Then measure where they sit relative to News Boxes and classic organic results. These are now different visibility layers.
• Audit which owned pages are actually being used. If Google is summarizing your topic, you need to know whether it is drawing from your current explainer, an old archive page, or a third-party site instead. BotRank’s GEO Page Analysis is useful here because technical accessibility and page clarity still influence what can be retrieved and reused.
• Refresh pages that answer durable questions clearly. The more a topic looks answerable, the more likely Google is to summarize it. Pages with clear headings, explicit facts, and current evidence are in a better position than vague editorial pages.
• Build a priority queue, not a wish list. Once you know where AI Overviews displaced older modules, turn that into concrete work by topic cluster, page type, and source gap. BotRank’s Recommendations feature is designed for that kind of execution layer.
• Watch link behavior, not just feature prevalence. If Google keeps adding richer link treatments inside AI answers, the winners will be the teams that can see whether those links lead to meaningful traffic or only cosmetic visibility. That is the practical follow-up to the link changes BotRank covered in its analysis of what SEOs should do next as Google AI search shows more links.

The key mindset shift is this: a lost News Box is not automatically a lost opportunity, and an AI Overview appearance is not automatically a win. What matters is whether your best evidence survives the transition into the answer layer, and whether the pages behind that answer are the ones you actually want Google to trust.

FAQ: what should SEO and publisher teams know next?

Does this mean Google will remove News Boxes entirely?

No one can say that as a fact yet. But the shift from 75% to 34% in News Box-only SERPs, alongside the rise in AI Overview-only SERPs, does make a smaller long-term role for standalone News Boxes look plausible.

Is this only a publisher issue?

No. Publishers feel it first because Top Stories is such a visible format, but any brand that depends on informational visibility should care. If Google is comfortable replacing a mature module in news-adjacent search, it can reshape other discovery patterns too.

Should teams still create content for queries that trigger AI Overviews?

Often yes, but with more discipline. If the page supports citations, brand trust, newsletters, apps, or other owned channels, it can still be worth producing even when CTR from web search gets weaker.

What is the first dashboard to build?

Build one view per topic cluster with four fields: AI Overview presence, position versus News Box, cited source mix, and downstream traffic. If your team cannot separate those layers, you will not know whether the problem is visibility, framing, attribution, or click-through.

Google’s answer layer is no longer a side feature. It is becoming the main surface users meet first. If your team still reports visibility as if the News Box and the blue links were the whole story, you are already measuring the wrong search results. The next step is to track the answer itself, audit the sources behind it, and decide where your content deserves to compete. If you want to do that with real prompts, source-level evidence, and an actionable optimization backlog, BotRank gives you the workflow to start now.