Microsoft Advertising says advertisers remain responsible for the AI-generated content in their campaigns, including when Microsoft’s own tools create it. Its guidance covers required disclosures, permission to use someone’s likeness or voice, and preserving information that identifies content as AI-generated.
The company’s dedicated help page, reported by Search Engine Land on September 14, addresses AI-generated, AI-manipulated and other synthetic advertising content.
It does not announce a universal requirement to put the same AI label on every ad. Microsoft ties mandatory disclosures to the laws that apply where an ad runs, with requirements varying by market and content type.
Microsoft’s watermark may be invisible to your customer
Microsoft says its AI tools apply metadata, machine-readable provenance information and imperceptible watermarks to generated images, audio and video. These signals identify how content was made, but may not appear as a label that a consumer can see.
Advertisers must still add any legally required visible or audible disclosure. Microsoft says it should be understandable and placed near the relevant content, and recommends embedding disclosures directly in image and video assets.
The distinction matters during production: a file can carry information identifying it as AI-generated while giving the person viewing the ad no visible indication of that fact. Advertisers need to preserve the former and check whether they must add the latter.
Microsoft also makes clear that disclosure does not excuse deception. Ads may be rejected, restricted or removed for prohibited deepfakes, impersonation, missing required disclosures, unauthorized use of someone’s likeness or voice, or interference with required machine-readable information.
The disclaimer tool can narrow your reach
The guidance points advertisers to Microsoft’s existing Disclaimer in ads feature, where the chosen format supports it. But that feature’s documentation contains a practical caveat: enabling a disclaimer means losing traffic from publishers that do not support disclaimers.
There are format restrictions too. The feature does not support Multimedia ads. For Audience campaigns, it works with Native ads. Performance Max support is limited to campaigns without a catalog feed serving Product ads.
Search advertisers can choose a popup or an additional line of text, with the latter replacing Ad text 2. Audience and Performance Max campaigns only offer the popup layout.
These are existing limits of the disclaimer feature, rather than a newly announced penalty for AI-generated ads. They mean advertisers considering the tool need to check both how the disclosure appears and which placements remain available. The AI guidance does not quantify any resulting traffic loss.
AI creation still needs human review
Microsoft already offers tools that generate ad text, images, backgrounds and videos. The transparency guidance leaves advertisers responsible for reviewing the resulting creative before submission and checking that its depictions and claims are accurate.
For campaign teams, that makes the final exported asset a useful review point: does the disclosure remain readable, has the file retained its provenance information, and does the ad accurately represent the product being sold?
SEW recently documented Google testing visible AI labels on generated review summaries in ads. Microsoft’s guidance addresses the advertiser’s side of that transparency problem: using an AI creation tool does not establish that the finished ad meets the disclosure rules where it will run.
Start the conversation by posting the first comment