Google AI Mode Search Updates: Latest Changes…
Google Search has changed more in the last twelve months than in the previous twelve years. If you have opened Google recently and found a conversational answer where a list of blue links used to be, you are not imagining it – and you are not alone in asking what happened to Google.
This page is a living record of Google AI Mode search updates, maintained by Szymaniak Digital. It answers the four questions marketing leaders actually ask: why Google has changed, how the new experience works, what has shipped and when, and what your business should do about it.
Written and maintained by Konrad Szymaniak, Founder of Szymaniak Digital, an enterprise SEO and AI search consultancy. Konrad has led search programmes for organisations including Frasers Group and Sports Direct, speaks at brightonSEO and MeasureFest, and guest lectures on enterprise AI search at the University of Southampton and the University of Stirling.
Last reviewed: August 2026
Google AI Mode Search Updates: What Happened?
Google spent 2026 rebuilding search around a conversation rather than a results page — and it did so in public, in roughly one significant release a fortnight.
The pivot point was Google I/O on 19 May 2026. There, Google confirmed that AI Mode had passed one billion monthly active users within a year of launch, with queries more than doubling every quarter. It made Gemini 3.5 Flash the default model in AI Mode for every user worldwide, and it shipped what Liz Reid, Google’s VP of Search, described as the biggest upgrade to the search box in more than twenty-five years. AI Overviews, which reached over 2.5 billion monthly users, and AI Mode stopped being two separate products and began running as one integrated flow.
Since then, the release cadence has not slowed. The most consequential Google AI Mode search updates of summer 2026:
- 12 June — Information agents launched. The first of the “Search agents” announced at I/O. Ask AI Mode to keep you updated on a topic and the agent carries on working after you close the tab, monitoring blogs, news, social posts and live finance, shopping and sports data, then pushing a summary with links. Exclusive to Google AI Ultra subscribers at launch, across every AI Mode market and language, with AI Pro access following later in the summer.
- 30 June — Recipe links moved to the top of responses. After sustained pressure from food publishers, AI Mode began surfacing prominent creator links with names, ratings and ingredient counts above the generated answer.
- 15 July — Personal Intelligence connected Google Calendar. Significant because Calendar is the first connected app AI Mode can write to, not merely read. It joins Gmail and Google Photos. US only at launch.
- 16 July — Third-party apps arrived. Instacart, Canva and YouTube Music can now be linked to AI Mode so tasks complete inside the conversation. US and English only for now.
- 21 July — Gemini 3.5 Flash-Lite began rolling out in Google Search. Google’s fastest, cheapest 3.5-class model, built for low-latency agentic work. Google has not specified what it powers inside Search.
- 22 July — France went live. The last major market to receive AI Mode and AI Overviews, two months inside Google’s own deadline, under commitments to a publisher opt-out, AI-traffic transparency and neighbouring-rights compensation covering roughly 450 French publishers and press agencies.
- 4 August — The measurement rules became clear. Google’s Search Console documentation confirmed that a follow-up question inside an AI Mode conversation counts as an entirely new query, with its own impressions, position and click data.
- 13–14 August — Gemini 3.7 Flash reached AI Mode. Added as a selectable model one day after release, globally, in English, for Google AI Pro and Ultra subscribers. Google says it follows instructions and reads intent more accurately. It is not the default.
And one important correction. In mid-July, several trade outlets reported that Google had “fully replaced” search with AI on 10 July, retiring the blue links worldwide. That framing overstates what Google has actually said. AI Mode is not the default Google Search experience. Google stated this at I/O 2026, and on 8 June 2026, after a Chrome Canary experiment routed address-bar searches into AI Mode, Google’s Search VP Rajan Patel went on the record to call it an error and to confirm there is no plan to make AI Mode the Chrome default.
The accurate picture is more useful than the dramatic one: AI Mode is a tab, an app entry point and an increasingly seamless handoff from AI Overviews — used by more than a billion people a month, but still sitting alongside a classic results page that has not gone anywhere.
Konrad Szymaniak, Szymaniak Digital: “The ‘Google killed the blue links’ headline is the single most expensive misreading in our industry right now. It sends boards into panic-mode budget decisions when the underlying reality is more nuanced and far more actionable. Classic organic results still carry the majority of Google’s volume. What has genuinely changed is that a second, differently-behaved surface now sits on top of them — and being visible on one tells you almost nothing about being visible on the other.”
And what about Google AI Mode Deep Search?
Deep Search is AI Mode’s research tier and remains reserved for Google AI Pro and Ultra subscribers. Where a standard AI Mode answer uses query fan-out — breaking your question into sub-questions and running them concurrently — Deep Search takes the same technique considerably further, issuing hundreds of background searches, reasoning across the findings and returning a fully cited, expert-level report. Expect two to three minutes of processing rather than an instant answer. It is built for the work that used to take an afternoon: market sizing, competitor analysis, due diligence, literature reviews.
For marketers, Deep Search is the surface where depth genuinely outperforms brevity. A 300-word summary page does not survive that level of scrutiny. A properly evidenced, well-structured resource does.
Google AI Mode Search Updates: Why Does It Matter?
Because the mechanics of visibility have changed underneath campaigns that are still being measured the old way.
1. Citation in AI Mode is far less durable than a ranking. Research published by SISTRIX in May 2026, tracking 82,619 prompts weekly across six countries, found that AI Mode cites between 14 and 16 domains per answer and replaces roughly 56% of them every single week. The churn is not evenly distributed: in 86.5% of prompts there is a stable core of one to five domains that persists for months, while around 89% of the domains outside that core rotate weekly. Encouragingly, in 83% of prompts at least one specialist domain sits in that stable core alongside the Wikipedias and YouTubes of the world. The core is not closed to smaller brands.
The strategic question is therefore no longer “do we appear in AI Mode?” It is “are we in the stable core, or in the weekly rotation?”
2. Winning on one Google surface does not win you the next. SISTRIX found AI Mode and AI Overviews citing entirely different domains on the same prompt 83% of the time. Ahrefs, measuring 540,000 query pairs, found the two surfaces citing the same URL just 13.7% of the time — despite the answers themselves agreeing in meaning about nine times out of ten. Each surface researches and writes independently.
3. In shopping, the shelf has become dramatically narrower. A Productrise study published on 29 July 2026, drawing on more than two million product listings across US and UK queries, found AI Mode surfacing around 95% fewer product listings than standard search. Two effects compound: standard search returned products on roughly 88% of tracked shopping queries against AI Mode’s 23%, and when products did appear, standard search averaged 22.5 per page against AI Mode’s 4.3. Overlap between the two surfaces was approximately 0.8%. The study counts listings rather than clicks, and covers product-centric keyword queries rather than conversational prompts — but the direction is unambiguous.
4. Your reporting cannot yet see what is happening. Search Console’s Generative AI performance reports, announced 3 June 2026 and still rolling out to a subset of site owners, report impressions only – no queries, no clicks, no click-through rate, no average position. Combine that with the rule confirmed in August, that every follow-up question generates a fresh set of impressions, and the risk becomes obvious: AI Mode impressions can be counted three or four times within a single conversation, with no click figure beside them. They measure how often Google used your page, not what that use was worth.
Blending generative and classic organic impressions into a single bucket, then calculating a blended CTR from the total, will produce a number that looks like performance data and behaves like noise.
5. Behaviour inside search has fundamentally shifted. Alphabet’s own disclosures put AI Mode queries at roughly three times the length of traditional searches, with about one in six arriving as voice or image rather than text, and daily queries per US user doubling in the product’s first year. Personal Intelligence adds another variable: with Gmail, Photos and Calendar connected, two people in the same city asking the identical question can now receive materially different answers.
Keyword-level thinking was already strained. Against three-times-longer, personalised, multimodal, multi-turn queries, it no longer describes the surface accurately.
Konrad Szymaniak, Szymaniak Digital: “Every enterprise client I work with is asking me to prove AI search ROI using a measurement stack that Google has not finished building. My honest answer is that we cannot yet – not to the standard a CFO would accept. What we can do is stop pretending impressions are traffic, instrument the things we can observe, and build the content assets that put us in the stable citation core rather than the weekly churn. The brands that do that quietly in 2026 will be unassailable in 2027.”
Google AI Mode Search Updates: Who Is Affected?
Every sector is affected, but not equally, and not in the same way.
E-commerce and retail – highest immediate exposure.
The 95% listing contraction and 0.8% overlap between surfaces mean strong Shopping visibility does not transfer into AI Mode. Product feed completeness now functions as a ranking input in its own right. Google’s Merchant Center AI performance insights, announced 27 May 2026, will eventually report share of voice across AI Mode, AI Overviews and the Gemini app for the US, Canada, Australia, India and New Zealand — but it has no confirmed ship date and no published methodology. Retailers are being asked to compete for a fraction of the slots while flying without instruments.
Publishers, media and content businesses – highest structural risk.
SISTRIX found news articles held a position in the stable citation core just 1.4% of the time – the weakest of any content type measured. Video was strongest at 24%. Meanwhile, regulatory pressure is producing real remedies: the UK CMA ordered a publisher opt-out from AI Overviews, AI Mode and Discover on 3 June 2026, with nine months to implement, and the French launch shipped with neighbouring-rights compensation attached. Recipe and food publishers have already extracted a visible concession in the form of prominent creator links.
Healthcare and regulated sectors – highest trust burden.
AI Mode synthesises rather than lists, which puts a premium on demonstrable clinical authorship and citable credentials. There is a live reputational dimension too: on 14 July 2026, Common Sense Media’s Youth AI Safety Institute rated Google Search’s AI features at its lowest risk tier for children, and Google has disputed the methodology. Healthcare brands should assume heightened scrutiny of the sources AI Mode chooses to trust — and structure their content to be obviously among them.
SaaS and B2B technology – highest opportunity.
Long, comparative, multi-part queries are precisely what AI Mode was built for, and precisely how software buyers research. Ahrefs found AI Mode names 2.5 times as many brand entities per response as AI Overviews — 3.3 against 1.3. Comparison pages, integration documentation, pricing transparency and genuine implementation detail are disproportionately rewarded. Deep Search, in particular, is a shortlist-building machine for enterprise procurement.
Financial services – highest precision requirement.
Conversational queries about rates, eligibility and products demand accuracy, current data and unambiguous compliance language. Personalisation raises the stakes further: an answer shaped by a user’s own Gmail and Calendar context is one a brand can neither see nor audit.
Enterprise and multi-location organisations – highest coordination.
AI Mode is live in over 200 countries and just under 100 languages, but features are not. Ads remain US-only. Connected apps are US-only. Side-by-side Chrome links are US-only. Information agents are Ultra-only. Agentic restaurant booking runs free in nine markets. Any enterprise reporting a single global “AI search performance” figure is averaging across markets with genuinely different products.
Google AI Mode Search Updates: What Should Businesses Do?
Six priorities, in the order we would sequence them for a client.
1. Separate your measurement before you interpret anything.
Split generative AI impressions from classic organic impressions in every report you produce. Never compute a blended CTR across the two. Document, in writing, that AI Mode impressions inflate with conversation length — so a rising impression count may reflect longer conversations rather than wider reach. Note also that Search Console’s chart data aggregates by property while the page table aggregates by URL, so page-level figures will not always sum to the chart total. That difference is a dimension change, not an error.
2. Audit whether you sit in the stable core or the weekly rotation.
Track your most commercially important prompts weekly, not monthly, and record which domains persist. This is the single highest-value diagnostic available today, and almost nobody is running it. Persistence, not presence, is the metric.
3. Build for Deep Search, not just for the snippet.
Deep Search issues hundreds of queries and produces a cited report. Thin content is invisible to it. Prioritise genuinely comprehensive, evidenced, well-structured resources on the topics where you need to be shortlisted — with original data, named expert authorship, clear entity markup and explicit publication and review dates.
4. Treat your product feed as ranking infrastructure.
For retailers, feed attribute completeness now determines whether an agent can evaluate and act on your products at all. Audit for gaps, enrich attributes, and structure product data so it can be read and acted upon by a machine, not only browsed by a human.
5. Get your technical foundations agent-ready.
Ensure critical content renders without JavaScript execution – agents inspecting your pages may simply miss anything that requires it. Confirm your nosnippet, max-snippet and noindex controls reflect a deliberate decision rather than an inherited default; these govern AI Mode exactly as they govern AI Overviews. Note that Google-Extended does not affect AI Mode at all — it governs Gemini model training only. If you want out of Google’s AI features entirely, the Search Console control introduced in June 2026 is the mechanism, and Google states it is not a ranking signal outside those features.
6. Report ranking and citation as two separate KPIs.
They are different outcomes, driven by different mechanics, and moving in different directions. A dashboard that shows only one of them is telling your board half a story.
What not to do: do not fragment your content into small chunks designed for AI consumption. Google’s Search Liaison, Danny Sullivan, stated in January 2026 that Google’s engineers specifically advise against it – Google’s systems parse full pages and extract relevant passages themselves. Equally, do not chase inauthentic brand mentions across forums and social platforms. Spam systems actively filter manipulation, and seeding fake mentions carries genuine downside risk to site-level trust signals.
Google AI Mode Search Updates: What We’re Watching Next
Will Gemini 3.7 Flash become the default? Google has not said. History suggests it will: Gemini 3 Pro arrived as an option in November 2025, Gemini 3 Flash became the default a month later, and Gemini 3.5 Flash replaced it at I/O in May. Flash-tier models remain Search’s production layer because of their latency and cost profile.
Google’s own support documentation still lists only “Fast” and “Pro” in the model menu, and has not been updated to reflect Auto or the new Flash tier – a small gap that tells you how quickly this is moving. Practical tip in the meantime: if you are seeing inconsistent AI Mode citations, run the identical query across 3.7 Flash, Auto and Pro and compare. You can now isolate whether the model itself is the variable.
- Search Console clicks.
The single most important unshipped feature in search marketing. Until Google reports AI Mode clicks as clicks, every ROI conversation about AI search rests on inference.
- Merchant Centre AI performance insights.
Announced in May, still not shipped, still without a published methodology. When it lands, retailers get their first Google-owned view of share of voice across AI surfaces.
- Search agents reaching the free tier.
Information agents launched Ultra-only in June with AI Pro access promised later in the summer. Every step down the subscription tiers materially expands a surface where your content can be surfaced without anyone having searched for you.
- Regulation as a product roadmap.
This is the underrated one. The UK CMA’s publisher opt-out carries a nine-month implementation window. The EU has ordered Google to share search data with rivals – AI chatbots included – with data flowing from January 2027. The EU’s July DMA fine explicitly flagged AI Overviews and AI Mode. The French terms — publisher opt-out, AI-traffic transparency, neighbouring-rights compensation — are the most publisher-favourable anywhere and are now a precedent every other press market can point at. Increasingly, what Google ships next in AI search will be shaped in Brussels and London as much as in Mountain View.
- Ads inside agentic flows.
Ads have been live in US AI Mode since late September 2025 and Alphabet still describes monetisation as early-stage, with a Direct Offers pilot testing merchant offers inside answers. As agents begin completing bookings and purchases without a website visit, the commercial architecture of that flow becomes the most important open question in paid and organic search alike.
Konrad Szymaniak, Szymaniak Digital: “My working forecast for the next twelve months is that AI Mode’s share of total search volume stays smaller than the headlines imply, while its influence on high-value, high-consideration queries grows faster than anyone is currently modelling. That asymmetry is the opportunity. If you optimise for volume you will conclude AI Mode does not matter yet. If you optimise for the queries that precede a purchase decision, you will conclude the opposite…. and you will be right.
Fifty-six per cent of the domains AI Mode cites are replaced every week. A small, stable core persists for months. Szymaniak Digital’s AI Search Visibility Audit tells you which side of that line your brand sits on, across AI Mode, AI Overviews, Deep Search and the wider generative search landscape — with a prioritised roadmap for staying in the core.
Frequently Asked Questions: Google AI Mode Search Updates
Is Google Search becoming AI search?
Partly, and not in the way the headlines suggest. Google Search is unquestionably shifting towards AI-driven, conversational answers: AI Overviews appear automatically above classic results and reach over 2.5 billion monthly users, AI Mode has passed one billion monthly users, and since I/O 2026 the two run as a single integrated flow. But AI Mode itself is not the default search experience. Google stated this at I/O 2026 and reconfirmed it in June 2026 regarding Chrome. Classic results remain the default; AI answers now sit on top of them and hand off into a deeper conversation when you want one.
How do I activate AI Mode on my Google Search?
In supported countries there is nothing to enable — it is already switched on. On desktop, use the AI Mode tab on any results page, go directly to google.com/aimode, or use the AI Mode option in Chrome’s address bar. On iPhone, open the Google app and tap the AI Mode icon beneath the search bar. On Android, use the Google app, the AI Mode button on Chrome’s new-tab page, or follow-up questions within Circle to Search. For voice and camera, Search Live is inside the Google app and has been available in every AI Mode market since March 2026.
Can I turn off AI in my Google searches?
There is no account-level setting to disable AI Mode, but because it is a separate tab rather than the default results page, you can simply choose not to use it. If what you want to avoid is AI answers inside normal results, that is AI Overviews, which is a different feature with different controls. Site owners — as distinct from searchers — have a separate option: the Search Console control introduced in June 2026 blocks a site’s content from Google’s AI features, and Google states it is not a ranking signal outside those features.
Is Google AI search free?
Core AI Mode is free with a Google account, and has been in over 200 countries and just under 100 languages. Agentic restaurant booking has been free in its nine markets since April 2026, and Personal Intelligence requires no subscription. Paid Google AI Pro and Ultra plans buy the frontier extras: Google AI Mode Deep Search, early access to new capabilities such as Search agents, higher daily limits on Gemini 3 Pro, and — since August 2026 — Gemini 3.7 Flash as a selectable model.
What happened to my AI on Google?
Almost always one of three things. First, feature availability differs sharply by market: connected apps, side-by-side Chrome links and ads are US-only, and information agents require an AI Ultra subscription. Second, some capabilities are English-only at launch, including Gemini 3.7 Flash. Third, AI Mode’s interface has been redesigned repeatedly through 2026 — the model menu now sits behind the “+” icon in the “Ask anything” bar, which is not where longstanding users expect to find it.
What is Google AI Mode Deep Search and how is it different?
Deep Search is AI Mode’s extended research mode, available to Google AI Pro and Ultra subscribers. A standard AI Mode answer uses query fan-out – decomposing your question into sub-questions and running them simultaneously. Deep Search takes that considerably further, issuing hundreds of background “grounded” searches, reasoning across disparate findings and producing a fully cited, report-length output. Allow two to three minutes rather than expecting an instant response. Use it for market research, competitive analysis, due diligence and any question where you would otherwise have spent hours.
When is Google going full AI?
Google has not announced any such plan, and has twice gone on the record in 2026 to say the opposite regarding defaults. The realistic trajectory is convergence rather than replacement: AI Overviews appear by default and hand off into AI Mode, AI Mode absorbs progressively more agentic capability, and classic results remain available beneath. For planning purposes, treat this as a permanent hybrid rather than a countdown to a switch-off date.
Why has Google Search changed so much?
Competitive pressure and capability, in roughly equal measure. ChatGPT and Perplexity demonstrated genuine demand for conversational answers, and Google – which still held around 91% of worldwide search in June 2026 – had both the index and the model capacity to respond inside the box where the overwhelming majority of searches already happen. Google’s own usage data reinforces the bet: AI Mode queries run about three times longer than traditional searches, roughly one in six arrive as voice or image, and daily queries per US user doubled in the product’s first year.
About Szymaniak Digital
Szymaniak Digital is a UK-based enterprise SEO and AI search consultancy helping senior enterprise marketing leaders stay visible as search becomes conversational, agentic and personalised.
Founded by Konrad Szymaniak, the consultancy specialises in enterprise SEO strategy, Generative Engine Optimisation (GEO) and Answer Engine Optimisation (AEO) – combining traditional organic performance with visibility across AI Mode, AI Overviews, Deep Search and the wider generative search landscape.
Konrad has delivered search programmes for organisations including Frasers Group and Sports Direct, holds an MSc from the University of Southampton, speaks regularly at brightonSEO and MeasureFest, and guest lectures on enterprise AI search and agent optimisation at the University of Southampton and the University of Stirling. He contributes to Semrush, Screaming Frog, Sitebulb, Majestic and Wordtracker.
Szymaniak Digital’s proprietary A.G.E.N.T. Framework and AI Recommendation Score give enterprise clients a measurable view of visibility across AI search surfaces – including whether their brand sits in the stable citation core or the weekly rotation.
Ready to find out where you stand? Book an AI Search Visibility Audit. Contact Us!
