On May 20, 2026, at Google Marketing Live, Google unveiled an entire new generation of ad formats built on Gemini. Conversational Discovery ads, Highlighted Answers in AI Mode, AI-powered Shopping ads in classic Search, Business Agent for Leads embedded inside the ad unit. Direct Offers expanded with native checkout and Gemini-built promotional bundles. It is not an update. It is a rebuild of the entire Google paid stack around AI. And the most important consequence for brands has nothing to do with ad performance – it has to do with who writes the words that describe them.
What Google actually announced
The new generation lands across four surfaces, all powered by Gemini.
Conversational Discovery ads generate creative dynamically in response to specific user questions. The advertiser supplies the product and the campaign signals; Gemini writes the ad copy tailored to the exact query – including, in Google's own example, a request to make a house "smell like those fancy spas or a rainy forest." The ad is not pre-written. It is synthesized.
Highlighted Answers place ads inside the recommendation lists that AI Mode generates – the "top apps for X" or "best products for Y" lists – framed as one of the answers, not as a sidebar unit. The ad sits where the editorial recommendation would, with a "Sponsored" label.
AI-powered Shopping ads bring the same Gemini-written approach to classic Search, not just AI Mode. When a shopper looks up an espresso machine, Gemini surfaces the most relevant products and "instantly writes a custom explainer" on why each one might fit. The advertiser writes none of the explainer.
Business Agent for Leads embeds a Gemini-powered chatbot directly inside the ad. A student looking at universities clicks "Chat" within the ad and has a conversation with an agent grounded on the advertiser's website. The lead form is gone – replaced by a dialogue.
On top of all four, Google added a transparency layer they call the "independent AI explainer": Gemini evaluates and synthesizes information about the product or service and displays its own commentary next to the advertiser's creative. The advertiser's claim sits alongside Gemini's framing of that claim, in the same ad unit.
And under all four, Google connected the dots to its own self-service tooling: to "take full advantage" of these formats, advertisers need to be on Performance Max, AI Max for Search, or AI Max for Shopping. The manual campaigns that thirty years of paid search rested on are now structurally disadvantaged.
The line that just collapsed
Until last week, the boundary between organic and paid AI visibility was clear. AI Overviews, AI Mode answers, and ChatGPT citations were the organic surface – probabilistic, content-driven, structured by what your site is and how the AI engines parse it. Google Ads, Shopping ads, and sponsored placements were the paid surface – auction-driven, copy-controlled, optimized by bidding strategy and ad rank.
Two different games. Two different teams inside the brand. Two different sets of metrics. The organic side worried about Citability Score and AI Visibility Index. The paid side worried about CPC, CTR, ROAS, Quality Score.
Last week, that boundary collapsed.
Inside AI Mode, the Highlighted Answer is shown as part of the AI's recommendation list. Inside Search, the AI-powered Shopping ad carries a Gemini-written explainer in the same visual language as an organic result. Inside both, the independent AI explainer is the system telling the user, in its own voice, whether the advertiser's claim is credible. The ad does not look like an ad. It looks like a continuation of the AI conversation, with a small "Sponsored" label that the eye learns to discount after the third encounter.
The brand no longer writes the ad. Gemini writes the ad, frames the ad, and explains the ad. The advertiser is the source material. The user reads a synthesis.
For the organic side, this is a familiar pattern. It is exactly how AI Overviews and ChatGPT citations have worked since 2024: the model reads what your brand has published, decides whether to mention it, decides how to frame it. The brand does not control the words. Citability Score was built to measure whether a brand was structurally prepared for that loss of control on the organic side. As of Google Marketing Live, the same loss of control now applies to paid.
Why the Citability Score now covers paid too
The mechanism that decides whether your AI-powered Shopping ad surfaces, what Gemini writes in its explainer, and how the independent AI commentary frames your product is the same mechanism that decides whether you appear in an AI Overview. It is a model reading structured signal about your brand and choosing what to say.
Citation Rate's 56 parameters that measure Citability Score include the kind of inputs that determine how an AI engine represents a brand: entity coherence, structured identity, semantic consistency between what the brand declares and what third parties confirm, technical functionality of the surfaces the model crawls, external reputation signals. These parameters were always meant to apply across AI engines. With Gemini ads, they now apply to the paid surface inside Google as well. A brand with a Citability Score of 35 going into Google Marketing Live 2026 will see Gemini producing weaker explainers, less compelling syntheses, and lower-credibility independent commentary than a brand with a Citability Score of 75. The bidding strategy will not save the lower-scoring brand. The AI will write what the structure allows it to write.
This is the structural shift that matters. Paid AI placement is no longer an auction problem. It is a citability problem with an auction layer on top. A brand can buy its way into the surface. It cannot buy its way into being well-described.
The independent AI explainer is a governance object
One of the four new formats deserves its own paragraph: the independent AI explainer. Google's framing is careful – the explainer "ensures transparency and builds trust" – but the practical reality is that a third-party system (Gemini) is now publishing commentary about a brand's product, inside an ad bought by that brand, framed as independent.
For a regulated brand – a bank, a pharmaceutical company, a regulated service provider in Europe – this is not just a creative challenge. It is a content-responsibility question. If Gemini's independent commentary about a financial product is incomplete, technically incorrect, or omits material disclosures, who is accountable? The advertiser, who paid for placement? Google, which generated the commentary? The AI Act, in its application to general-purpose AI systems, has begun to address questions like these, but the field is moving faster than the regulatory framework.
For now, the operational consequence is clear: brands operating in regulated sectors will need to audit what Gemini says about them inside Google's own ad units. Not as a creative review, but as a compliance review. The output of an AI ad is no longer a static asset that legal signs off on once. It is a generated artifact that changes per query. Sampling, monitoring, and evidence retention become part of the workflow.
This is exactly where AI visibility measurement starts to overlap with AI governance, and where the conversation moves from marketing to risk. We will return to this in a separate piece.
What this changes operationally
Three practical shifts follow from the Google Marketing Live announcements.
The organic-paid wall inside the brand needs to come down. The team that worked on AI visibility (or didn't) and the team that worked on Google Ads (or did) have been operating in separate dashboards, separate budgets, separate KPIs. With Gemini-written ads, the inputs that determine organic citation quality are the same inputs that determine paid ad copy quality. A brand that has invested in Citability Score now sees a return on the paid side as well. A brand that has neglected it now sees the cost on the paid side too. Internal silos that survive this transition will produce worse paid campaigns, not just worse organic visibility.
The "ad creative" as an asset class is changing shape. For years, the creative was a finished artifact: copy, image, video, landing page. With Conversational Discovery ads and AI-powered Shopping ads, the creative is generated per query from underlying signals about the product. The work shifts from writing the copy to structuring the inputs the AI will use to write the copy: product feed quality, entity-level descriptions, brand voice signals available across the open web, structured data on the site. The creative team converges with the data team converges with the citability team.
Measurement needs to extend to paid AI output, not just paid AI performance. Click-through and conversion will still matter. But a new measurement axis becomes operational: what is Gemini actually writing about my brand in its independent explainers, across how many query variations, with what variance, with what factual accuracy. This is not a Google Ads question. It is an AI Visibility question applied to a paid surface. The same instruments that monitor citation on the organic side need to extend to monitor synthesis on the paid side.
The bigger pattern, confirmed
In a piece last week, the argument was that Google is in a controlled tapering of the classic SERP – constrained only by the advertising revenue base that requires the classic interface to stay alive long enough to migrate. Google Marketing Live 2026 is exactly that migration in motion. The advertisers are being moved, deliberately and at scale, from a SERP-anchored auction model to an AI-anchored synthesis model. Performance Max, AI Max for Search, and AI Max for Shopping are the migration vehicles. The new ad formats are the new destination.
The advertising revenue stream that anchored Google's transition is now flowing through the AI interface itself. The classic SERP will still exist for years, generating residual ad revenue while the migration completes. But the structural center of the Google ads business has moved. And with it, every advertiser that wants to stay competitive has to move too.
What to do this week
Three practical checks worth doing before the new formats roll out broadly. First, run a brand audit on what Gemini currently writes about your products when prompted with realistic query variations – the output you see today is approximately what the independent AI explainer will produce in the new ad formats. Second, look at your Citability Score and the structured inputs feeding it: product feed completeness, entity descriptions, third-party validation signals. The same inputs will now determine paid ad quality, not just organic citation rate. Third, if you operate in a regulated sector, raise the question internally now: who reviews AI-generated ad output for compliance, and what evidence chain do we keep. The new ad formats turn into live products in coming months. The governance work needs to happen before, not after.
Google did not just announce new ad formats. It announced that the last untouched part of the brand-to-AI relationship – the part where the brand still wrote its own words – is now also AI-mediated. The line between organic citation and paid placement was the last wall between two different ways of being represented. That wall is gone.