Google Ads Now Writes Your Product Ads. Stay in Control

At Google Marketing Live 2026, Google handed Gemini the pen for your product ads — AI Shopping ads that read your Merchant Center feed and write a custom explainer for each shopper’s query, with AI Max matching now core to Search. For advertisers, the shift isn’t whether the machine writes the ad; it’s whether you still steer what it says, where it shows, and how you measure it. This guide maps the new controls — and which ones you have to switch on yourself.

Two dates frame the moment. AI Max for Search — the consolidated matching-and-creative layer that succeeds broad match — reached general availability on April 15, 2026. A month later, at GML 2026 on May 20, Google previewed AI-written Shopping ads and extended AI Max to Shopping and Travel. Automation is no longer a setting you opt into; it is becoming the default surface your budget runs on.

Below: what actually shipped versus what is still rolling out, the six advertiser controls that keep you in the driver’s seat, the branded-search toggle nobody at Google has formally announced, and a shifting migration calendar that rewards teams who keep re-checking it. Where a number comes from Google’s own marketing, we say so — which is why Digital Applied’s paid media team treats these as governance settings, not fire-and-forget defaults.

The headline from GML 2026 is that generation is moving inside the auction. In the AI-powered Shopping format Google previewed, Gemini reads a retailer’s Merchant Center product feed and — per query, not per campaign — writes a custom explainer describing why a specific product fits what the shopper asked for. Google’s own example was a category search for “espresso machines,” where the system surfaces your most relevant product and explains the match in its own words.

That is a real change in who authors the ad. But read the rollout language carefully before you rebuild your strategy around it — and note that AI Max query matching, the other half of the handoff, is already the live default on Search.

AI Max for Search is not a new campaign type. It is a consolidated feature set that layers on top of your standard Search campaigns, announced by Brandon Ervin, Director of Product Management at Google Ads, when it reached general availability on April 15, 2026. It bundles three capabilities — and each one has an off-switch or a guardrail, which is the part most recaps skip. If you want the fuller agency-side breakdown of the GA milestone and the DSA sunset, we covered it in our AI Max general-availability and DSA-sunset playbook.

Alongside the AI-written Shopping format, Google announced AI Max for Shopping at GML 2026 — a distinct upgrade that extends AI Max’s text customization, Final URL Expansion, and automatic format selection (text ad versus Shopping ad, whichever better matches intent) to Shopping campaigns. It reads Merchant Center feed attributes — Google’s example cites “fabric softness, material durability and fit” — to answer the conversational, long-tail queries that standard Shopping campaigns miss. Existing advertisers can upgrade in one click, and Google announced a parallel AI Max for Travel on the same framework.

The dependency is uncomfortable but clarifying: when Gemini writes from your feed, feed quality stops being a data-hygiene chore and becomes your creative brief. Thin or generic attributes produce thin, generic explainers. That is why getting your product feed ready for AI-generated Shopping ads and preparing merchant product data for AI shopping surfaces now sit upstream of the campaign itself, not downstream of it.

Here is the part no single announcement pulls together. As Google automates matching and creative, the levers that used to be campaign-level tactics become governance settings — and they are scattered across a Google blog post, the Help Center, and a trade-press scoop. The matrix below maps all six advertiser-facing controls, what each one governs, where in the pipeline it is enforced, and — critically — whether it does anything before you turn it on.

Read the enforcement column and a pattern emerges. Your hardest controls sit earliest in the pipeline: negative keywords are enforced before AI Max expansion, so any query containing one is excluded no matter how relevant the model judges it. The softer levers — AI Brief, Text Guidelines — steer generation rather than block it, and the branded-search toggle is not even confirmed. The takeaway for 2026: govern at the gate you can trust, and treat the generative controls as influence, not guarantees. For the messaging side specifically, our brand-safe text guidelines framework for AI Max turns brand voice into concrete term and messaging restrictions.

The most consequential control for anyone protecting brand traffic is also the least documented. A branded-search control has been spotted inside AI Max offering three settings: show ads on all relevant searches (the default), apply inclusion or exclusion rules for branded queries, or restrict delivery to unbranded searches only — using Google’s own brand-entity index rather than a list you maintain. The rationale is straightforward: advertisers running a separate branded Search campaign don’t want AI Max cannibalizing those clicks at a higher cost and muddying attribution.

The catch is provenance. This was not a formal Google blog announcement. It was first surfaced in limited rollout by paid-search practitioners on LinkedIn in late May 2026 and covered by trade press in early June, and it was not visible in every account at the time. Presented with the same confidence as the AI Max GA date, it would mislead — so we flag it plainly as reportedly rolling out.

Google’s automation timelines are not fixed points — they move, which is itself an argument for governance over set-and-forget. Google originally set a single September 2026 auto-upgrade date for Dynamic Search Ads, Automatically Created Assets, and the campaign-level broad-match setting. On June 11, 2026, Google Ads Liaison Ginny Marvin posted a Developer Blog update delaying the DSA automigration specifically to February 2027 — while ACA and campaign-level broad match still auto-upgrade in September 2026 as planned. Blanket-stating “DSA delayed to February 2027” without that carve-out misstates what actually changed; we unpack the reprieve in our DSA-to-AI-Max delay playbook.

Demand Gen is where much of the retired inventory now lands, and it picked up its own 2026 changes — a Commerce Media Suite and YouTube view-through optimization in April, then automatic creative resizing, Gemini creative recommendations, and a Web-to-App acquisition report in a June 25 update. Google attributes acquisition value to the channel with two of its own statistics: it says campaigns that include TV screens drive an average of about 7% additional conversions at the same ROAS, and that 72% of incremental YouTube conversions come from new customers. Both are Google-sourced marketing figures with no visible independent methodology — directional, not audited. If Demand Gen is on your roadmap, our Demand Gen 2026 advertiser playbook covers the billing and format shifts in depth.

One more signal sits outside the advertiser product line but inside the same governance climate: effective June 15, 2026, Google removed the browser back-button as a trigger for full-screen vignette interstitials across AdSense and Ad Manager, citing its policy against back-button hijacking. It is a publisher-side change, not a Shopping or Search feature — but it fits the broader 2026 pattern of Google tightening manipulative-UX rules on one surface while expanding automation on another.

The through-line of every change above is the same: control did not disappear, it moved into governance you have to configure. Here is how we sequence it for clients — inputs first, guardrails next, brand protection, then measurement that lives outside the platform selling the automation.

None of this is a one-time setup. The levers, defaults, and migration dates all keep moving, so the controls are recurring reviews, not a launch checklist. Teams that operationalize them keep more of the upside from automation while capping the downside; teams that accept the defaults inherit whatever Google optimizes for. If you want a second set of hands on the governance layer, our paid media service runs these as ongoing programs, and we tie them to independent analytics and measurement so the incrementality story doesn’t come only from the platform being measured.