
On November 19, 2026, Microsoft Advertising made its AI content transparency controls available to every advertiser, so you can now declare whether an image or video was made with AI. Where the ad format supports it, you can also ask for a visible watermark, the "Made with AI" style mark that viewers can see on the ad.
If your Microsoft ads use product photos, lifestyle scenes or short videos that were generated or edited with AI tools, this applies to you. Microsoft says you decide whether your content needs a label, and you stay responsible for each declaration you make.
The first job is a short inventory: list the images and videos in your live ads, then match each one to one of three declaration values. The sections below cover where to do it, what each value means, and how to check the finished ad.
Key Takeaways
Microsoft's declaration field has three values (AIWithVisibleWatermark, AIContent and NoAIContent), and you choose the one that matches each image or video.
The visible watermark appears only where the format supports it, and its size and placement vary, so preview the rendered ad in your account.
Microsoft's detection signals are a reference, a label does not fix a misleading image, and a saved record of each choice protects you.
Microsoft now lets you declare AI-made images and videos in your ads
Microsoft's announcement on its Advertising blog says the controls "help you declare AI-generated image and video content and manage visible labels across supported ad formats." In practice, you tell Microsoft whether an image or video in your ad includes AI-generated content, and you can request a visible watermark where the format supports one.
You can manage declarations for one asset or many at a time in the Ads Studio Library, through the Microsoft Advertising APIs (the connections that let software change your account), and through product feeds (the files that list your products for shopping ads). The supported formats are Product Ads, Multimedia Ads, Audience Ads, Display Ads, Video Ads, and assets inside Performance Max campaigns, where you upload the images and videos and Microsoft assembles the ads.
Microsoft's announcement says you decide whether your content needs a label and you remain responsible for your own declarations. It adds that governments are passing laws that may require labels on some AI-generated content, and that you are responsible for making sure your ads follow local laws and disclosure rules. For a legal question about your market, ask a lawyer who handles advertising.

Three declaration values cover every image you run
Microsoft's product documentation describes the AI label as "the advertiser's declaration about AI-generated or AI-modified content," and calls it "an advertiser declaration, not a system-generated AI detection signal." The field takes one of three values.
| Value | What you are telling Microsoft | Visible watermark |
|---|---|---|
| AIWithVisibleWatermark | The asset is AI-generated or AI-modified | Requested where supported |
| AIContent | The asset is AI-generated or AI-modified | Not requested |
| NoAIContent | The offer contains no AI-generated or AI-modified content | Nothing to request |
A requested watermark is a request, not proof that one exists. Microsoft's documentation says the watermark value asks for a watermark where supported; it does not certify that one is already on the asset. Microsoft's AI detection signals, its own indicators of whether content looks AI-generated, are described as a helpful reference. You still make the call.

A five-step check tells you which value fits each asset
You can do this inventory in one sitting with access to your Microsoft Advertising account and whoever made your images. Microsoft's help page on AI transparency and its API guide cover the exact settings.
The five-step asset check
Open your active campaigns and the Ads Studio Library, and write down every image and video in use, including Performance Max assets.
Note which assets were generated or substantially edited with AI tools, and ask the designer or agency when you cannot remember.
Microsoft's guidance asks advertisers to preserve watermarks, metadata and provenance information, so export from the original tool, not a stripped copy.
Choose AIWithVisibleWatermark, AIContent or NoAIContent for each asset.
Write one line per asset with the value, who made it, and the date you decided.
Pay extra attention to any asset that shows real people, your actual products, a real location, an endorsement or an event. Compare what the image shows with what the ad claims. A generated scene of "our team on a job site" is a different matter from a generated background behind a real product, and the record you keep should say which one it is.

The label's look depends on the format, so preview the ad first
Microsoft says watermark size and placement vary by ad format, placement and serving environment. A watermark request that looks right on a Display ad may look different on a Product Ad, and the same asset can appear in several places.
So, before an ad runs, preview each format that uses the asset in your account and record what the rendered result looks like. Then check the finished ad for fit: the label should not cover your product, your price or your phone number in the preview.
After the ads are live, keep judging them by customers, not labels. Our posts on seeing which Bing ads turn into customers and on Microsoft's bidding upgrade cover that side of the account.
Google Ads and Meta ask for AI labels in their own ways
If you advertise on several platforms, each one has its own control and its own wording, so one decision on Microsoft does not carry over. Google Ads says you can add labels directly to your creatives or use AI label settings in Google Ads, Display & Video 360, Campaign Manager 360, Merchant Center and Ads Editor. Google adds that labels may be required in some regions and for some content types, and that its setting does not guarantee compliance with local rules.
Meta works differently. Its announcement on generative AI transparency says ads created or significantly edited with Meta's own generative AI tools may receive an "AI info" label. In a June 1, 2026 update, Meta says it is also beginning to detect third-party AI signals and place "AI info" inside "About this ad."
| Platform | Where you act | What viewers may see |
|---|---|---|
| Microsoft Advertising | Declare in the Ads Studio Library, APIs or product feeds | A visible watermark where supported |
| Google Ads | Add a label to the creative or use AI label settings | Labels where used; Google-generated images carry machine-readable markup and SynthID |
| Meta | Ads made or significantly edited with Meta's AI tools may be labeled | "AI info" in "About this ad," and next to "Sponsored" for some photorealistic AI-generated people |
Keep one asset record with a column for each platform. If you generate images with Google's tools, our post on making ad images with Google's AI image model shows how that workflow runs.
Every AI image still has to tell the truth
Microsoft says a label does not make deceptive content acceptable, and its guidance asks advertisers to preserve provenance information, the data stored with a file that records where it came from, and to provide disclosures where applicable rules call for them. Google likewise tells advertisers to review AI-generated assets for accuracy and misleading implications before publishing.
The FTC's advertising guidance for small businesses says ads must be truthful, non-deceptive and backed by evidence. A generated product photo that shows a finish, size or result your product does not have is a problem no declaration solves. Replace the image, then declare the new one correctly.
A simple house rule keeps every declaration consistent
A rule that works for small accounts: if AI generated the asset or substantially edited it, declare it as AI content. If nobody can say how an image was made, treat it as unknown: ask the person who made it, or replace it with an asset whose history you can state. Choose NoAIContent only when you know the offer contains no AI-generated or AI-modified content.
Then decide, once, whether your brand wants a visible watermark requested on AI-made assets or none, and apply that choice the same way across campaigns. That consistency is part of how branding holds together when AI tools produce some of your visuals. Our post on using AI images in business marketing covers the wider questions about rights and customer trust.

Can you pick the right AI declaration?
Pick an answer to begin.
1. Which value tells Microsoft an image is AI-generated and requests a visible watermark where supported?
2. You pick AIWithVisibleWatermark for a Display ad. What does that guarantee?
3. Who decides whether a piece of content needs an AI declaration in Microsoft Advertising?
Frequently Asked Questions About Microsoft Ads AI Labels
Where do I set an AI declaration in Microsoft Advertising?
Microsoft lists three places: the Ads Studio Library, the Microsoft Advertising APIs and product feeds. The library lets you manage declarations for one asset or many at a time, and the other two suit larger accounts and shopping feeds.
What is the difference between AIContent and AIWithVisibleWatermark?
Both say the asset is AI-generated or AI-modified. AIWithVisibleWatermark also requests a visible watermark where the format supports one, while AIContent does not request a watermark.
Does Microsoft's AI detection pick the declaration for me?
Microsoft describes its AI detection signals as a helpful reference. The declaration is yours to make, and you remain responsible for it.
Which ad formats can carry a declaration?
Microsoft lists Product Ads, Multimedia Ads, Audience Ads, Display Ads, Video Ads and assets in Performance Max. Watermark size and placement vary by format, placement and serving environment.
Can I add a label to fix an image that overstates my product?
A label does not make deceptive content acceptable, according to Microsoft. Replace or correct the image so it shows what the customer will receive, then declare the new asset.
Do Google and Meta use the same system as Microsoft?
Each platform has its own controls. Google offers AI label settings and adds markup and SynthID to images its tools generate, and Meta uses its "AI info" label. Keep a separate column for each platform in your asset record.
Moving Forward
Microsoft's AI content transparency controls give you three declaration values, three places to set them, and a request for a visible watermark where the format supports one. The work is an inventory: list the live assets, mark the AI ones, keep the metadata, choose the matching value, and save a record. Then preview the rendered ad, because the label's look depends on the format.
Done once and kept up, that record makes every later campaign faster. New images get the same question before they run, and a change of designer, agency or tool does not leave you guessing about how an old asset was made.
If you would like help, Web Leveling can review your live Microsoft ad assets, set up the declarations and keep a record that covers each platform. Our pay-per-click work includes that kind of asset review. We work with small and medium businesses across the country and overseas. Tell us what your ads currently use, and we will help you decide which declaration fits each image.
Terms
AI label words in this post
Tap a term to see what it means.
Declaration. Your statement to Microsoft about whether an image or video includes AI-generated or AI-modified content.
Visible watermark. A mark on the ad that viewers can see, requested through the declaration where the ad format supports it.
AI detection signal. Microsoft's own indicator about whether content looks AI-generated, offered as a reference while you decide.
Provenance information. Data stored with a file that records where it came from and how it was made.
Product feed. A file that lists your products and their details, including images, for shopping ads.
Performance Max. A campaign type where you supply assets such as images and videos and the platform assembles the ads.




