The EU AI Act Now Requires Disclosure for AI-Generated Ad Creative

Since August 2, the EU AI Act's transparency rules under Article 50 have been legally binding, and they reach further into advertising than most marketers realize. If your ads use AI-generated or AI-manipulated visuals, and reach an EU audience, you now have a legal disclosure obligation unless a human has meaningfully reviewed and taken editorial responsibility for the content. This isn't a distant compliance deadline anymore, it's already in force, and it lands right on top of a year where AI-generated ad creative went from a novelty to the default across Google and Meta.

What Actually Requires Disclosure

The obligation targets AI-generated or AI-manipulated content that resembles real people, objects, or events closely enough to be mistaken for authentic, commonly described as deepfake-adjacent content, when it's used to inform the public on matters of public interest, a category that legal guidance explicitly extends to certain advertising, PR, and corporate communications. In practice, this covers product shots altered to look different from the real product, digital replicas or likenesses of real people, and de-aging or similar manipulation effects applied to people in ad creative. Straightforward stock-style AI image generation without a real-world reference point sits in murkier territory, but anything that could pass as an unaltered photo or video of a real person or product is squarely inside the rule.

Why This Collides With What's Already Live on Ad Platforms

Meta's Advantage+ Creative now rewrites text baked into your ad images by default, and Google is auto-generating missing video aspect ratios for Performance Max using generative AI, both changes we've covered as they rolled out this year. Neither platform's default automation currently distinguishes between "safe to leave unlabeled" and "requires disclosure under EU law", that judgment call sits with the advertiser, not the tool generating the asset. An ad running EU-wide with AI-extended video content or an AI-altered product image can now be non-compliant even if the advertiser never manually asked for anything, simply because a platform's default automation touched the file.

What to Actually Do About It

• Audit which of your live creative assets have been touched by platform-side AI generation. Advantage+ Creative rewrites, Performance Max video reframing, and any other automated AI editing on your assets are now compliance-relevant, not just performance-relevant, for anything reaching EU audiences.
• Use the official EU disclosure icons rather than improvising your own label. The Code of Practice provides three standardized icons specifically for this purpose, and using the recognized format is the safer path versus a custom disclaimer that may not satisfy the requirement.
• Establish a human review step before EU-facing creative ships, not after. Meaningful editorial review is the actual exemption path, a documented review process for AI-touched creative gives you a defensible position if a specific asset is ever questioned.

A transparency law that's already in force, colliding with ad platform defaults that generate and modify creative automatically, is exactly the kind of compliance gap that doesn't show up until a specific ad gets flagged, and by then it's a legal question, not a settings question. If you run creative that reaches EU audiences and want your AI-touched assets audited against this requirement, our paid media team can review your current setup before it becomes a compliance issue instead of a settings toggle.