Table of Contents
- What Google Changed on November 5, 2026
- Why Shopping Merchants Should Care
- What Counts as AI-Created or AI-Edited
- How the AI Label Setting Works in Practice
- Where This Collides With Product Feed Rules
- Google's Own Automatic Improvements
- A 30-Minute AI Asset Audit
- If You Are Suspended, Do Not Blame the Label
- Three Realistic Scenarios
- A 90-Day Plan for AI Disclosure Readiness
- Frequently Asked Questions
What Google Changed on November 5, 2026
Search Engine Roundtable and several other trade outlets reported on November 5, 2026 that Google expanded the list of regions where advertisers must disclose the use of AI in ads. California, South Korea and Vietnam were added to a list that already included the European Union, India and New York. Google made the change in its own help documentation for the AI content label setting, and the same language was reported to have been added to the Merchant Center version of the page.
The underlying rule is straightforward. Where AI regulations require it, ads that use assets created or edited with generative AI must carry a disclosure. Advertisers can either build the label into the creative itself or use the AI label setting that Google began rolling out gradually in July 2026. When an asset is marked as AI-created or edited, Google adds a note in the "How this ad was made" panel and, for campaigns targeting the listed regions, a visible overlay on the ad.
Region list and setting behavior are taken from Google's Advertising Policies Help page on AI labeling updates (July 2026) and trade coverage dated November 5-6, 2026. Google controls the final wording, so confirm current details in the help center before you rely on any single sentence here.
Why Shopping Merchants Should Care
It is tempting to assume this is a creative-team problem for brands running video or Demand Gen. It is not. Merchant Center sellers increasingly use AI to generate lifestyle product images, remove or replace backgrounds, extend images to new aspect ratios, write titles and descriptions, and build short product videos. Some of that happens in third-party tools, some inside Google's own automatic improvements, and some in Performance Max asset generation. Every one of those touches the question of whether an asset was "created or edited with AI".
Merchant Center also sits at the intersection of two sets of rules: the product data and image policies, and the advertising policies that govern how those assets show up in ads. An AI-generated image that misrepresents the product, or one with a watermark or text overlay, can violate image requirements long before any AI label question arises. If you have not read our guide to the relaxed image quality rule on promotional overlays, do that first, because the AI label overlay is a separate thing from the overlays merchants are not allowed to add themselves.
What Counts as AI-Created or AI-Edited
Google does not hand out a neat legal definition for every case, so use a conservative working definition. An asset is probably AI-created or edited if a generative model produced any visible part of it that was not in the original photograph or recording. The list below separates the clear cases from the gray ones.
- Clearly yes: a fully synthetic product photo, a generated model wearing your garment, a generated background or scene around a real product, a synthetic voice-over, a generated person in a video.
- Probably yes: generative fill that extends a canvas or removes objects with visible changes, AI upscaling that alters texture, AI-generated variations of a real product shot.
- Usually no: ordinary retouching such as color correction, cropping, exposure adjustments and conventional background removal that does not add content.
- Ask your own lawyer: anything where the output could be mistaken for a real photograph of a real product that does not look like that in real life.
The risk is asymmetric. Disclosing something that did not need a label costs you a small overlay. Failing to disclose something that did need one can create a policy problem in regulated regions, plus a trust problem with shoppers.
How the AI Label Setting Works in Practice
According to Google's help page, advertisers can designate assets as AI-created or edited, and labelled assets then receive the disclosure note in the three-dot menu on the ad. In campaigns that target the listed regions, the ad also carries a visible overlay. Google says the setting launched gradually, so you may not see it in every account yet.
- Inventory every asset type you upload: product images, additional images, lifestyle images, video links, logos, headlines, descriptions.
- Flag which of those used a generative tool at any stage, including tools that bundle AI quietly, such as editors with one-click background generators.
- In accounts where the AI label setting is available, mark those assets as AI-created or edited rather than leaving the default.
- Check which regions your campaigns target. If you serve the EU, India, New York, California, South Korea or Vietnam, assume the visible overlay can appear.
- Re-run the check whenever your creative team or agency changes tools.
Note that Thailand was reportedly not on the list, and the list is likely to grow. Treat the region set as a moving target and build the habit of labelling at source rather than reacting region by region.
Where This Collides With Product Feed Rules
Disclosure does not make a bad image acceptable. A generated image that shows features your product does not have is still a misrepresentation issue, label or no label. Google's image guidelines still prohibit watermarks, promotional text, and images that are not an accurate representation of the product. See our image requirements guide for the baseline and the 2027 image size enforcement timeline for resolution rules coming into force.
A good rule for stores selling physical goods: the primary image_link should be a real photograph of the actual product. Use AI for supplementary imagery only, and only if the generated scene does not alter the product itself, its color, its size cues, or its included accessories. That policy protects you in a Google review, in a chargeback dispute, and under consumer protection law.
A generated image that shows a different color, finish, or set of accessories from what you ship will draw complaints and returns. It can also contribute to the product-data and misrepresentation signals described in our guide to misrepresentation versus product data violations.
Google's Own Automatic Improvements
Another source of AI-touched imagery is Merchant Center's automatic improvements. If you have enabled image improvements, Google may adjust backgrounds or crop images for you. Our article on the Automations page explains what those settings do. The key point: automated edits made by Google are Google's responsibility to label, but you are responsible for deciding whether to switch them on and for reviewing the output.
Review a sample of your approved products in Shopping search results at least once a month. Look for odd crops, removed accessories, or clean-up artifacts that change how the product looks.
A 30-Minute AI Asset Audit
Run this before the holiday peak
List every AI tool anyone on your team or agency uses for images, video, copy or voice.
Export your top 50 products by revenue and check each primary image is an unaltered or conventionally retouched photograph.
Identify generated lifestyle images and confirm they do not misstate size, color or contents.
Mark AI-created or edited assets in the AI label setting where it is available.
Confirm your target regions in each campaign and note which fall under on-ad overlay rules.
Write a one-paragraph internal AI content policy so freelancers and agencies follow the same rules.
Document the date and result of the audit, so you can show a reviewer your process if asked.
If You Are Suspended, Do Not Blame the Label
No source we reviewed says that a missing AI label alone triggers a Merchant Center account suspension. In practice, suspensions follow patterns such as misleading images, inconsistent shipping or return information, and missing business details. An AI image problem usually shows up as an image-quality or misrepresentation issue. If you are already dealing with a suspension, our suspension recovery checklist is the right starting point, and the appeal writing guide covers how to explain fixes to reviewers.
Three Realistic Scenarios
Scenario 1: an apparel store using generated models. The store replaced studio shoots with AI-generated models wearing real garments. The garment is photographed, the model is synthetic. Under a conservative reading this is AI-created content, so the assets should be marked, and the store should confirm that fit, drape and color are faithfully shown. Returns data is the early warning: if returns citing 'not as pictured' climb, the images are overpromising. Scenario 2: a home goods store with generative backgrounds. The product photo is real but the room around it is generated. The product is unchanged, yet the asset is still AI-edited. Mark it, keep the original studio shot as the primary image, and use the lifestyle image as a supplement. Scenario 3: an agency running Performance Max for twenty clients. The agency's creative tool generates asset variations by default. Nobody remembers which clients' assets included generated content. The fix is process: a per-client asset register that records the tool, date, and whether an AI label was applied. Our PMax asset groups guide describes how assets are organized, which makes the register easier to build.In all three cases the common thread is traceability. If you can show when an asset was created, with which tool, and what was marked, you can answer any reviewer, regulator or customer question quickly.
A 90-Day Plan for AI Disclosure Readiness
Regulations around AI disclosure are moving quickly, and Google has already expanded its list twice since July. Rather than reacting to each announcement, build a lightweight program.
- Days 1 to 15: audit current assets and tools, as in the checklist above, and write your internal policy.
- Days 16 to 45: mark existing AI-created or edited assets wherever the setting is available, and replace primary images that are synthetic with real photographs.
- Days 46 to 75: train freelancers and agencies, add an AI disclosure field to your creative brief template, and require suppliers to state whether their assets used generative tools.
- Days 76 to 90: review Google's help pages for changes in regions or definitions, and re-audit a sample of live assets.
Put a recurring calendar reminder to read Google's AI labeling help page once a quarter. A five-minute read can save you from discovering a new region only after complaints arrive. If you sell internationally, also check the advertising rules of the destination country, since local laws may impose obligations beyond Google's list.
Frequently Asked Questions
Does the AI label rule apply to all Shopping ads? No. The visible overlay applies to campaigns that target the listed regions and use assets marked as AI-created or edited. The note in the ad details menu applies more broadly. Confirm current scope in Google's help page.
Do I need to label AI-written product titles and descriptions? Google's examples focus on creative assets. Text generated with AI is a gray area, so check the current help documentation and apply the same honesty test: is every claim accurate and verifiable?
Can I use AI to create my main product image? Technically you can try, but it is a poor idea. Use a real photograph for the primary image to avoid misrepresentation and returns problems, and use AI only for supplementary imagery.
Will this change affect my Merchant Center approval? Not directly according to the sources we reviewed. The risk is indirect: poor or misleading AI imagery can violate image and misrepresentation policies.
Where do I find Google's official wording? Search the Advertising Policies Help center for the AI labeling requirements page and the Merchant Center Help page for AI-generated content. Both are updated over time.
Check the Basics Before the Overlay Does
AI labels are one more place where your assets, feed and site need to agree. A free scan at gmcunbanned.com checks your store for the policy and landing page issues that most often trigger suspensions.
Run Free GMC Scan →