The Ad Ran and Nobody Approved It

An ad is a promise, and sooner or later someone checks it against the real thing. If the ad claims something the brand cannot deliver, that is where it breaks. Which is why it matters that a Google Ads headline can serve under a brand's name before anyone at the brand has read it. In the roughly seventeen years I built and led a creative team at Full Sail University, approval meant a person stood behind the specific work going out the door.

Beginning September 2026, Google is automatically upgrading Search campaigns that use campaign-level broad match or standalone automatically created assets to AI Max, with Dynamic Search Ads following in early 2027.

What actually changes if Google Ads has been writing ad copy for years?

Machine-written ad copy is not new. Google Ads has generated headlines from website content through Dynamic Search Ads and automatically created assets for years, both features an advertiser controlled. What changes under AI Max is the default: every new Search campaign starts with text customization and final URL expansion switched on, and pinned assets stop holding whenever a more relevant page gets picked instead.

The migration mirrors the legacy setup, so a campaign-level broad match campaign without automatically created assets gains no machine-written copy from it. The change shows up in the next campaign built. Say a headline is pinned because the exact wording matters. With text customization and final URL expansion on, Google may swap in a more relevant page and generate copy to match, and the pinned asset is not used. The guardrail is still in the interface. It just stops holding.

Why does reviewing an ad after it runs not count as approval?

Generated assets in Google Ads are eligible to serve when generated, and a brand team cannot create or edit one, only remove it. Reviewing the asset report can come after the ad has run under your name, which makes it a check on the record, not a step before anything goes out the door.

Google rechecks assets at least every 48 hours and refreshes them as needed, including when a landing page changes. Change the page and the ad can change with it. So there is no fixed version to approve. A review that clears today's assets says nothing about the next refresh.

I came up in production as a producer, director, and editor. A final cut gets signed by one person who watched the specific frame and can stand behind what it shows. Google says it checks generated assets with quality models for accuracy and grounding in your landing page content. A grounding check is not a signature. No one on your team signs the specific promise before it runs under your brand's name.

What does approval look like when your landing pages become ad copy?

Google Ads lists your domain, landing pages, existing ads, and ad group keywords as generation context, so a brand's domain is source material for ad copy as well as the place a click lands. Approval has to move upstream to what a brand team actually controls: the pages on its own domain, how a campaign is structured, and the text guidelines applied to it.

Every claim on every page has to be defensible standing alone. "Move-in ready" is true of the building that is delivered and false of the phase still under construction, and a model drawing on your whole domain can carry the phrase from one page into an ad for the other. A page that oversells one floor plan can become a headline that oversells the whole community.

Campaign structure is where the hard switch lives: turn text customization off and generated assets stop serving, final URL expansion turns off with it, and a pin holds unless URL inclusions are set. Ad groups shape what the model draws from but guarantee nothing. A claim that must stay exactly as written belongs in a campaign with text customization off, and account-level automated assets and auto-applied ad recommendations need to be off as well.

Text guidelines are the third lever, still in beta and only available with text customization on: term exclusions and messaging restrictions for text customization assets, set per campaign. Google says they need to be direct requirements. They cut both ways. Exclude "move-in ready" and it disappears from every generated asset in that campaign, including the ones for the building that is ready. The real fix belongs on the page.

An algorithm does not flag the edge cases. It finds them and serves them if it predicts they will perform.

Who is responsible when Google Ads writes a headline that is not true?

You are. Google tells advertisers to keep website content accurate and to monitor generated ad content for accuracy, which leaves that job with the brand. Authorship can shift to Google Ads under AI Max. Responsibility for what the ad claims stays with whoever owns the domain the copy was drawn from.

The rules that govern what an ad can claim do not relax because a model wrote the line instead of a person.

Approval stops being a signature on the finished ad and becomes the work done before there is an ad, in pages nobody thinks of as ad copy and settings nobody thinks of as creative work. It is less visible on the calendar and much harder to get right.