The "was this ad made by AI?" question now has an official answer
Quietly, over a couple of weeks in July, the two companies that sell most of the world's digital advertising made the same call: AI-made ads now come with receipts.
Google has started automatically labeling ads created with its generative-AI tools. The disclosure lives in the "How this ad was made" section of My Ad Center, across Search, YouTube, and Discover. Meta moved in lockstep — ads built with its AI tools get auto-labeled, and advertisers using outside AI tools have to disclose it, with the tag surfacing in each ad's "About this ad" menu.
Put together, AI disclosure just became the default condition of digital advertising. Not a voluntary gesture. The baseline.
How the labeling actually works
Google's system doesn't depend on advertisers being honest. Detection runs on SynthID — an imperceptible watermark baked into the output of Google's generative tools — plus C2PA metadata signals that extend labeling to AI content made elsewhere. For external tools that leave no watermark, Google is adding a self-declaration control.
The architecture is the point. Watermark detection means the label follows the asset, not the paperwork. An AI-generated image stays labeled even after it's passed through three agencies on its way to the platform.
The deadline behind the virtue
Don't mistake this for a sudden attack of conscience. Google's rollout landed 24 days before an EU regulatory deadline, and the requirement notably covers all ads — not just election advertising, where synthetic-media rules usually stop. Transparency arrived on a compliance schedule.
Motive aside, the effect is the same. And platform rules built for the EU rarely stay in the EU — it's simpler to label everywhere than to fork your ad stack by geography.
What changes for marketers
Three practical shifts. First, inventory your stack. If generative tools touch your creative — Google's, Meta's, anyone's — assume a label will follow it, and check where third-party tools now require self-declaration. The compliance risk lives in the gap between what your team used and what you disclosed.
Second, the creative math changes. A visible "AI was used here" tag is now part of the ad experience. Some audiences will read it as neutral plumbing; others, as a reason to keep scrolling. Expect brands to A/B test not just creative but how disclosure interacts with trust. The winners will treat the label as a design constraint, not a scarlet letter.
Third, provenance joins the media-buying checklist. As C2PA metadata spreads through creative tools, proving how an asset was made — human, AI, hybrid — becomes part of the supply chain, the way brand-safety verification did before it.
The bigger shift
Advertising has run this play before: sponsored-content labels, political-ad archives, tracking disclosures. Each started as regulation-driven friction and ended up as furniture. AI labels are next in line.
The real story is the plumbing underneath. Watermarking and provenance metadata are now wired into the world's two biggest ad systems, and that infrastructure won't stay confined to ads. "Was this made by AI?" is getting a machine-readable answer — advertising is just where it's being installed first.
Image: Florian Wehde, via Unsplash
