By the Mayab Media Team
Six places now. That's how many regions Google Ads lists where an ad built with AI-created or AI-edited assets can carry a visible label on the ad itself.
Back in July the list was the European Union, India and New York. On November 5, Search Engine Roundtable spotted that Google's help page had quietly grown to include California, South Korea and Vietnam. No blog post. No email to advertisers that we've seen. Just an edited help document and a matching line in the Merchant Center product data spec, which says AI rules in those six places require disclosures or labels on ads with certain AI-made assets.
If none of your campaigns reach those six places, you can stop reading. Everyone else, keep going. A US campaign set to target the whole country includes California, whether you thought about it or not.
What the label actually does
Google gives you a switch on each asset. You mark an image, video or other asset as created or edited with AI, or you mark it as not. When a campaign targets one of the listed regions and uses an asset you've flagged, Google puts an overlay on the ad that viewers can see.
There's a second, quieter layer too. Easy to miss. Ads on Search, YouTube and Discover have a three-dot menu that opens My Ad Center, and its "How this ad was made" section shows the disclosure. That part has been around since the summer.
And if you made the asset inside Google's own AI tools? Google already embeds SynthID watermarks and C2PA data in those images and videos, and any label Google applies itself can't be overwritten by you. So the real decisions sit with the stuff you made elsewhere: the Midjourney hero image, the product shot your designer cleaned up with generative fill, the video your agency stitched together with an AI voiceover.
The part Google won't promise
Here's the line in the help doc we'd underline twice: using the label setting doesn't guarantee you comply with any particular law. Google tells advertisers to get legal advice and take whatever steps they need.
Read that plainly. Google built the toggle. It didn't sign up to be your compliance department. If a regulator decides your ad needed a disclosure, "I didn't flip it" or even "I flipped it" may not be the end of the conversation. The doc also gives no effective dates for any region and says nothing about penalties for skipping the label. We couldn't confirm either from Google's pages, so don't treat silence as a free pass.
We're not lawyers, and we won't pretend to know how California, South Korea or Vietnam will enforce their rules. What we can tell you is how to make sure your account isn't the weak link.
What we'd do in your account this week
Start with an honest inventory. Ask whoever makes your creative which assets were generated or edited with AI tools. Not "mostly real photos with a bit of touch-up." Which ones, specifically.
Then label them. It's tedious work. In Google Ads, go to Tools, then Asset studio, then Asset library. Pick the asset, choose the option to label it as created or edited with AI (or the one that says don't label), and save. When you build a new campaign, Google may also show a Review assets prompt for ads that might need a label. Don't click past it.
A few more things worth checking:
- If you add your own label in Photoshop or Canva instead, keep it away from the very edges and corners of the image. Google's auto-cropping can slice it off.
- Turn off Asset Enhancements, and Adaptive Layouts in Demand Gen, for any asset carrying a homemade label, for the same reason.
- In Merchant Center, keep the IPTC metadata on AI-generated product images. Google's spec says it must be there, so don't let an image optimizer strip it.
- If AI wrote your product titles or descriptions, the spec has fields for that: structured_title and structured_description, each with a digital_source_type value.
Is it worth the hour? Some math. Say you run a home goods store spending $8,000 a month on Shopping and Performance Max, and 40% of that lands in California. That's $3,200 a month riding on ads that may now need a disclosure. An hour of cleanup is cheap insurance next to a disapproval or a legal letter.
Will the label hurt performance?
Maybe a little. Nobody has published solid data yet, and we'd be suspicious of anyone claiming they have. Our guess: for most product and service ads, a small AI tag won't move click-through much. For brands that sell trust (clinics, law firms, financial advisors), an AI badge on a smiling stock-style face might cost you more than you'd like.
That's an argument for real photography in those categories. It isn't an argument for hiding AI use. Marking assets honestly is the cheaper path, and Barry Schwartz at Search Engine Roundtable expects this list to keep growing, which we'd bet on too.
Know which of your ads were made with AI before a regulator asks you.
Want us to check how this affects your business? Book a free audit.
