Microsoft Advertising Just Set the Rules for AI-Generated Ads

Microsoft Advertising has published formal guidance on AI-generated, AI-manipulated, and other synthetic content used in ads. The message is direct: every disclosure, consent, and accuracy requirement stays in force whether AI built the creative or a person did. If anything, the bar sits higher now.

Microsoft is drawing a clear line: using AI to make an ad is fine. Using AI to mislead the person seeing it gets the ad rejected. The guidance gives marketers a concrete set of obligations to work through before their next AI-assisted campaign goes live.

What Microsoft actually requires

The guidance covers three areas that every advertiser running AI-assisted creative now needs to own.

Disclosure. Where disclosure is legally required, it must sit close to the content it refers to. Microsoft recommends embedding AI disclosures into the image and video assets themselves, and its existing ad disclaimer feature supports this on compatible formats.

Provenance. Advertisers are expected to preserve watermarks, metadata, and other provenance signals that identify how a piece of AI-generated content was created. Stripping that information out counts the same as omitting a required disclosure.

Compliance by market. Rules on disclosure, consent, and the use of a person’s likeness or voice vary by jurisdiction. Microsoft holds advertisers responsible for meeting those requirements in every market where a campaign runs.

A label only goes so far

Microsoft’s guidance draws a firm boundary here. Deceptive creative stays deceptive, however clearly it carries an “AI-generated” tag. Ads can still be rejected, restricted, or removed if they contain prohibited deepfakes, impersonate a real person or organisation, use someone’s likeness or voice without proper authorisation, skip a required disclosure, or interfere with machine-readable provenance data.

Content produced with Microsoft’s own AI tools can carry machine-readable provenance data and imperceptible watermarks. These signals often stay invisible to a consumer, so an advertiser may still need to add a visible or audible disclosure on top of what the platform embeds.

What this means for marketing teams

This guidance turns AI ad transparency into a process to manage, with clear steps for marketing teams to own.

• Build a review step for AI-assisted creative before submission, checking that people, products, places, claims, and events depicted are accurate
• Treat watermarks and metadata as a fixed part of the asset that no production or editing step should touch
• Check disclosure requirements per market before a campaign goes live
• Brief creative and legal teams together on AI-assisted work, since disclosure and provenance now sit on both sides of that line

The wider signal

Microsoft is among the first major ad platforms to put this level of structure around AI-generated advertising, with more likely to follow. Advertisers who build disclosure and provenance checks into their workflow now stay ready as other platforms introduce similar requirements. Those who treat this as a formality to work around are the ones most likely to see their ads rejected, restricted, or pulled.