The checkbox appears at the top of the ad-creation screen, one line above the budget field: “Created or edited with AI.” From July 9, Google will not run a commercial ad on its largest properties unless an advertiser ticks it. The change is quiet, a small form tweak, but it rewrites the rules for the tens of billions of dollars in advertising that flow through Google Search, YouTube and the Discover feed each year.
Google announced on July 9 that it is extending its generative-AI disclosure requirements to all commercial advertising across those three surfaces. The stated purpose is consumer protection: viewers should not be fooled by synthetic scenes, by a product demo that never happened, by a testimonial delivered in a voice that belongs to no one, by a resort rendered entirely in software.
“We want people to have the context they need to judge what they are looking at,” a Google spokesperson said in the announcement.
The scope is broad. Ads generated entirely by AI must carry the label, as must ads that contain AI-created or AI-edited elements, including synthetic faces, generated voices, simulated environments and altered footage of real events. Ads that use AI only for production mechanics, such as resizing an image or correcting color balance, are exempt.
The regime itself is not new. Google already required AI labels on election ads and on certain political content, and it has required creators to label realistic synthetic content on YouTube. What changed on July 9 is reach: the requirement now applies to all commercial advertising across Search, YouTube and Discover, and the company says it will enforce it through the same machinery it uses for other ad policies, meaning automated scanning, manual audits and suspension for repeat offenders.
The timing is deliberate. Regulators on both sides of the Atlantic have been circling synthetic content for two years. The European Union’s Digital Services Act already obliges platforms to label deepfakes, and the U.S. Federal Trade Commission finalized a rule in May banning AI impersonation of real people in commercial settings, with a broader labeling framework under review. Google’s move, made by announcement rather than under formal order, lets the company set its own standard before a regulator sets one for it.
Advertisers have little room to object. “If a label is the price of admission to YouTube inventory, brands will pay it,” said a media buyer who works with major consumer brands, speaking on condition of anonymity. “The open question is whether the label changes anything about how people watch the ad.” Early research on synthetic-content labels has been mixed: some studies find that labels lower trust in the ad itself, while others find that viewers stop noticing them once they become routine.
The label may do its most important work on the economics of AI-generated creative. Agencies have begun producing speculative campaigns with generative tools, rendering dozens of variations of a single spot and letting optimization algorithms pick the winner. Under the new rule, every variation carries the same disclosure, which dulls the novelty that made synthetic creative attractive in the first place. A campaign engineered to look authentic now starts with a visible asterisk, and agencies said the asterisk changes how they pitch AI-first work to clients.
YouTube creators face a parallel obligation. Google has long required realistic synthetic content on the platform to be labeled; the July 9 rule extends the same principle to the advertisers who buy the space between videos, which means the label will now sit on both sides of the commercial break.
Industry analysts said the policy is likely to spread. Microsoft, Amazon and Apple each run ad networks with their own content rules, and each has been drafting internal AI-disclosure language. Once a company of Google’s scale makes the label a default condition of media buying, a competitor that skips the label carries the risk of being caught running unlabeled synthetic ads.
Enforcement is where the policy will live or die. Google said it will combine automated detection, which scans creative for signs of synthetic generation, with manual review, and advertisers will be required to self-certify at the campaign level. The company has also been a quiet backer of content-provenance standards, the technical specifications that let a piece of media carry a machine-readable record of how it was made, and industry executives said provenance metadata will increasingly be the evidence behind the checkbox. A label without underlying proof, one ad-technology executive said, is just a claim; the provenance record is what makes it auditable.
Google’s incentives pull in two directions. The company sells billions of dollars in ads against inventory that increasingly includes AI-generated imagery; too strict a rule could slow the supply of creative that keeps its ad system profitable. Too loose a rule invites regulators to act. The July 9 choice, broad scope with a light process, a single checkbox at the top of the screen, is Google’s attempt to split that difference.
Whether one checkbox changes consumer trust remains an open question. What is settled is the form change itself, now in force across the three biggest surfaces in digital advertising, and advertisers will learn the answer in the coming quarters.
