On July 9, 2026, Google published a changelog entry titled “Updates to AI labeling requirements (July 2026)” and quietly changed how every Detroit advertiser has to think about creative production. The entry introduced the AI label setting — a control that rolled out gradually through July across five products: Google Ads, Display & Video 360, Campaign Manager 360, Merchant Center, and Google Ads Editor. If your Michigan business has been using Asset Studio, Product Studio, or any generative tool to spin up image and video assets, some of your ads are about to carry a disclosure whether you flip the switch or not.
This is not a soft recommendation. Google’s documentation names three circumstances under which it applies a label without advertiser input, and in each case states plainly that those labels “cannot be overwritten.” For agencies and in-house teams across Metro Detroit running lean creative operations on AI-assisted production, this is the moment to audit the asset library — before a compliance question arrives from a client, a legal team, or a regulator.
What is the Google Ads AI label setting?
The AI label setting is a per-asset control that lets you declare whether a given image or video was created or edited with generative AI. Once declared, that disclosure surfaces in two distinct layers, and understanding the difference matters more than any button you’ll click.
The first layer is global. Any ad containing an asset flagged as AI-created or AI-edited carries a disclosure inside the “How this ad was made” panel in My Ad Center — the consumer-facing screen reachable through the three-dot menu on ads across Search, YouTube, and Discover. A customer in Royal Oak and a customer in Bangalore can both click through and read the same underlying disclosure. Nothing about the ad’s surface changes.
The second layer is regional and far more visible. For campaigns targeting the European Union, India, or New York State, flagged assets also trigger a visible overlay directly on the creative. The identical ad looks different depending on where it serves. Michigan has no equivalent statute on the books, which is exactly the trap: a Detroit manufacturer or a Southfield services firm running a national campaign that includes New York in its geo targeting will see labeled creative in that market and unlabeled creative in Wayne County, from the same asset, in the same campaign.
Why did Google roll this out in July 2026?
Regulation, and a deadline. Rules in the European Union, India, and New York require that ads containing certain AI-generated or AI-edited assets carry disclosures. Google shipped the changelog entry on July 9 with EU enforcement beginning August 2, 2026 — roughly three weeks of runway. The company was not being generous with lead time; it was clearing its own liability.
The rollout also did not appear from nowhere. The trail runs back several years:
- February 19, 2024 — Google requires an IPTC metadata tag on AI-generated product images in Merchant Center, a narrow early version of the rule.
- May 21, 2025 — YouTube’s mandatory AI content disclosure for creators takes effect, establishing the automatic, non-adjustable labeling logic now reused on the ads side.
- December 18, 2025 — Google adds SynthID verification to the Gemini app, letting anyone check whether a video carries the watermark.
- June 24, 2026 — Google Ads API v24.2 ships synthetic-content labeling structures at the programmatic level.
- July 9, 2026 — The AI label setting arrives across five products.
Google is explicit that using its setting does not guarantee compliance with any specific law. Its documentation pushes advertisers toward legal counsel. That caveat is worth repeating to any Michigan client who assumes flipping a toggle ends the conversation.
Which labels can you not remove?
Three scenarios trigger an automatic, permanent label:
- Google is legally required to apply one in a given region.
- Google receives signals from another platform indicating AI involvement.
- The advertiser used one of Google’s fully automated tools to generate and serve the creative.
That third condition is the one that catches small and mid-sized Detroit advertisers. If your team generated a background, an image variant, or a product shot inside Asset Studio or Product Studio because it was faster than booking a photographer in Eastern Market, the resulting asset carries a label you cannot strip. There is no interface control to remove it.
Underneath the visible layer, Google embeds SynthID — a non-visible digital watermark — plus markup conforming to the C2PA standard from the Coalition for Content Provenance and Authenticity, which attaches cryptographically signed metadata describing an asset’s creation and editing history. Both are woven into assets generated inside Google Ads tools automatically, independent of whatever the visible setting says. The metadata lives inside the file. Stripping the interface label does nothing to it.
How does this affect YouTube ads and video assets?
YouTube is where the two disclosure regimes collide, and where Michigan advertisers running video are most likely to get tangled.
Uploading a video to YouTube directly from inside Google Ads triggers a separate prompt asking whether AI was used. That question is tied to YouTube’s organic content disclosure policy — the one that took effect May 21, 2025 — not the ads-side AI label setting. They are two different systems asking similar questions, and answering one does not answer the other.
If you select an existing YouTube video that already carries a label created outside Google Ads, the interface offers a link back to wherever that label originated so you can review it rather than resetting it. In Display & Video 360, YouTube video assets tied to ads must be labeled through an SDF file or the API after upload, with SDF-based labeling available from version 10.1 onward.
For a Detroit dealership group or a Michigan healthcare system running Demand Gen and video reach campaigns with AI-assisted b-roll, the practical consequence is that a single video may need its disclosure status set in three places: YouTube Studio, Google Ads, and whichever DSP is trafficking it.
How do you turn on AI labels in Google Ads?
Inside Google Ads, “AI label” appears as a column in both the asset library and asset reporting, and you can filter assets by that status. Three workflows exist depending on where you are in the campaign lifecycle:
- During campaign creation — a “Review assets” prompt appears when ads may require a label. Select the assets, choose Manage AI label, pick “Label this asset as created or edited with AI” or “Don’t label this asset,” then select Done.
- For existing campaigns — go to Tools > Asset studio > Asset library, open an individual asset’s details page, apply the label choice, and Save.
- When adding new AI-generated assets — a table lists assets with the “AI label” column undesignated; checking a box surfaces the same labeling pop-up.
In Merchant Center, the control sits under “Manage images” in a product’s details, inside the three-dot menu at the top right of the asset. In Campaign Manager 360, an AI labels dropdown appears after you save a creative upload. In Display & Video 360, the Creatives gallery filter bar breaks results into three states — on, off, and not set — and bulk edits run roughly ten seconds per creative, so a large batch takes several minutes with the browser window open. In Ads Editor, the shared asset library exposes an “AI label” column with a pencil icon for single or bulk edits.
One cross-engine caveat worth flagging for anyone on Search Ads 360: AI label fields work for Google Ads and Microsoft assets, but assets from other connected engines such as Yahoo Japan or Baidu default to “Unknown” and stay there, because Search Ads 360 has no way to push the label through.
Can you build your own AI disclosure label instead?
Yes. Google’s documentation explicitly states that a custom disclosure label built in your own design tools will not violate its separate policies banning text overlays and watermarks — provided the label functions as a disclosure rather than as decorative or misleading text.
Two constraints follow from how Google processes uploaded images. First, ad rendering can trim the outer edge of a creative, so Google’s guidance is to not place labels within a 5.5 percent margin around the full perimeter of the asset. Second, Google’s automatic image-enhancement features crop and adjust assets to fit placements, and can crop a custom label out of frame without warning. Opting out avoids the risk: the control is called Asset Enhancements in Display and Performance Max campaigns, and Adaptive Layouts in Demand Gen.
What should Detroit and Michigan advertisers do this week?
Michigan has not passed an AI advertising disclosure law, but geography does not protect you here. Detroit’s advertiser base skews toward multi-state and cross-border campaigns — automotive suppliers selling into Ontario and the EU, logistics firms targeting the New York corridor, healthcare and higher-ed institutions recruiting nationally. Any of those targeting profiles pulls you into a labeled jurisdiction.
- Run an asset library audit. Filter by the “AI label” column and find every asset sitting in the “not set” state. Unknowns are the liability.
- Check your geo targeting against the three regions. If EU, India, or New York appear anywhere in your campaign structure, visible on-ad overlays are live in those markets.
- Reconcile status across products. An asset labeled in Google Ads does not inherit that status in Display & Video 360 or Campaign Manager 360 if uploaded separately. Each product keeps its own field. This reconciliation is the bulk of the real compliance work.
- Test creative performance with and against the label. IAB research published in January 2026 found a wide gap between how advertisers assume consumers feel about AI-generated ads and how consumers actually report feeling — Gen Z trust dropped 19 points. Assume the label is a variable, and measure it.
- Keep election ads on their own track. With Michigan’s 2026 statewide races in full swing, note that election advertisers must still disclose synthetic or digitally altered content via the “Altered or synthetic content” checkbox in campaign settings, under Google’s Political content policy. That obligation is separate from and unchanged by the new AI label setting.
The bigger shift: disclosure moves from optional to enforced
The meaningful change here is not a new column in the asset library. It is that disclosure has moved from something an advertiser might volunteer to something enforced at the interface level in three markets, sometimes automatically, with SynthID and C2PA metadata running underneath as a check that survives even when a person downstream strips the visible label.
Detroit teams that treat this as a paperwork exercise will discover it too late, in an ad that renders differently than the mockup they approved. Teams that treat it as a creative-strategy question — deciding deliberately which assets are AI-generated, which are human-shot, and what the label does to click-through in each market — will end up with cleaner accounts and better answers when a client asks. If you want help auditing your Google Ads and YouTube asset libraries before the labeling reconciliation gets ahead of you, the team at Rosh Media works with Michigan businesses on exactly this kind of platform-change triage.
The five-product rollout finishes this month. EU enforcement starts August 2, 2026 — days away. The asset library audit is a two-hour job today and a fire drill in September.
Comments are closed