Google Ads AI Labels: Where C2PA Actually Sits

On July 9, 2026, Google switched on "How this ad was made," a disclosure panel now reachable worldwide in My Ad Center, covering ads on Search, YouTube, and Discover. "Reachable" is the operative word: the panel itself is accessible everywhere, but a visible label stamped on the ad itself is only mandatory in a handful of places, the EU, India, and New York. Everywhere else, nothing shows unless a viewer clicks in. The press release frames all of this entirely around SynthID, Google DeepMind's invisible watermark. What it does not mention is that Google's own support pages describe a second layer underneath: C2PA Content Credentials , the same open provenance standard this site covers on cameras, on YouTube, and under the EU AI Act. That omission matters, because the two layers do not cover the same ads.

What is "How this ad was made"?

It is a click-through disclosure panel attached to individual ads. Viewers open it from the three-dot menu or the info icon on an ad, and it shows whether the ad's creative was made with generative AI. It applies across Google's largest ad surfaces: Search, YouTube, and Discover.

Google splits how that disclosure gets attached into two paths, and the difference between them is the actual story here.

Does Google use C2PA for ad transparency?

Yes, but only for ads made with Google's own generative AI ad tools. Google's support page for advertisers states this directly, in a section titled "Note about machine-readable metadata (SynthID and C2PA)":

A second support page, for the My Ad Center panel itself, repeats the same point: "we provide detailed provenance information in the form of post-click label disclosures in My Ads Center as well as SynthID and C2PA to integrate machine-readable metadata." Both pages link out to Google's own C2PA transparency post , the same commitment Google and OpenAI made in May 2026 when they aligned on shipping C2PA and SynthID together.

Source: support.google.com/My-Ad-Center-Help/answer/17196133

None of this is new territory for Google. It already applies C2PA elsewhere, in Search and Chrome. What is new is the extension into ad creative specifically. The blog.google announcement simply never says the word C2PA, so a reader who only sees the press release would assume Google invented a standalone proprietary system for ads. It did not. It reused the standard.

The catch: third-party AI tools get no C2PA at all

The C2PA and SynthID embedding described above applies only to assets generated with Google's own ad tools. For ads built with a third-party generative AI tool, such as Canva, Adobe Firefly, or a Midjourney-class model, the advertiser instead flips a manual disclosure control declaring the ad was AI-made.

Google does not independently verify that self-declaration. There is no cryptographic backing, no manifest, no watermark check. It is an honor system: the advertiser says the ad used AI, and the disclosure panel reflects whatever the advertiser entered.

Where do the labels actually appear?

Disclosure visibility varies by region. In the EU, India, and New York, Google can show an AI label directly overlaid on the ad itself, per the regional visible-label requirements described in the same support documentation. A viewer sees the disclosure without clicking anything.

Everywhere else, the disclosure lives one click deep. It does not appear on the ad by default. A viewer has to know to open the three-dot menu or info icon and find "How this ad was made" to see it at all. That gap between regions is worth knowing if you are evaluating how much real-world visibility this feature delivers outside the jurisdictions that require it.

How does this line up with the EU AI Act deadline?

The EU AI Act's Article 50 transparency obligations , covering AI-interaction disclosure, machine-readable synthetic-content marking, and deepfake labeling, take effect on August 2, 2026. Those obligations were not touched by the "Digital Omnibus on AI," the provisional political agreement reached May 7, 2026 that delayed the separate Annex III high-risk system rules to December 2, 2027 and the Annex I product-embedded rules to August 2, 2028. Article 50 stayed on its original schedule.

Google's ad-transparency rollout landed on July 9, 2026, about three weeks ahead of that Article 50 deadline. Advertisers running Google Ads campaigns that reach EU audiences now have a working disclosure mechanism in place before the legal requirement actually bites. That timing looks deliberate. Our guide to Article 50 and Content Credentials covers what the regulation actually requires. The mechanism follows the same pattern we saw when YouTube started reading C2PA metadata to auto-label AI videos in May 2026. Video got the C2PA treatment first. Ad creative just got folded into the same standard, not a new one Google built from scratch.

How to check an ad's C2PA manifest yourself

Google's support docs make a specific claim: ads built with its own generative AI tools carry an embedded C2PA manifest. That claim is checkable. It does not require trusting the disclosure panel at all.

1. Save the ad's creative. Save the image or, for video, the file itself, or copy the direct media URL if the ad links to a hosted asset.
2. Open the inspector. Drop the file into C2PA Viewer , or paste the URL using the "From URL" option. Verification runs locally in your browser. Nothing is uploaded to a server.
3. Read the manifest. If a signed C2PA manifest is present, the inspector shows the generating tool, timestamps, and any AI-generation claim. See our guide on how to verify C2PA content for what each field means.
4. Treat a missing manifest as informative. No manifest does not automatically mean the ad is not AI-generated. It means the creative did not come through a Google-tool path that embeds one, or the file was re-encoded somewhere along the way and stripped it. Either way, you now know the disclosure panel's claim rests on the advertiser's word, not on cryptographic proof.

This is a small check, but it is the difference between reading a label and verifying a claim. Given that Google itself draws a hard line between its own-tool path and the third-party honor system, checking the manifest is the only way to know which side of that line a specific ad actually falls on.

Frequently Asked Questions