AI Content Labeling Is Here. Brands Need a Plan.
In 2024, California tried to pass AB-3211, a bill that would have required AI providers to watermark synthetic content and social platforms to label it. The bill died in the Senate before becoming law. Two years later, the underlying idea is now enforceable regulation, just not from Sacramento.
The EU AI Act's Article 50 went into effect on August 2, 2026, requiring providers of generative AI systems to embed machine-readable marks in synthetic audio, image, video, and text. OpenAI, Google, Meta, Microsoft, and Anthropic have all signed the accompanying Code of Practice. AI content labeling is no longer hypothetical. For brands using generative AI in creative work, the question has shifted from "will this happen?" to "what does this mean for us?"
The short answer: less than you might think, and more than you might hope.
The Duty Falls on Providers, Not Brands
This is the part most coverage gets wrong. Article 50(2) of the EU AI Act places the machine-readable marking obligation on the providers of generative AI systems, meaning Adobe, OpenAI, Google, and every other tool in your creative stack. Not on the agencies or brands using them.
If your team used Firefly to extend a background in Photoshop, Adobe is responsible for embedding the provenance mark. If you generated copy with an AI writing tool, that provider carries the marking duty. Brands are downstream of the obligation, not the source of it.
That said, brands are not entirely off the hook. Article 50(4) creates a separate visible-label duty for deployers, which includes anyone publishing content. Specifically:
- Deepfake-style content depicting real people, places, or events must carry a visible disclosure at first exposure
- AI-generated text on matters of public interest must be labeled unless it has undergone genuine, documented human editorial review
For most brand advertising, neither condition applies. A product image with an AI-extended background is not a deepfake. A social post about a sale is not a matter of public interest. The labeling requirement is narrower than the headlines suggest.
The Transparency Argument for Brands
Even where labeling is not legally required, there is a strategic case for leaning into transparency rather than away from it.
Brands have long included "authentic" in their values. The instinct when AI labels arrive is to worry that "synthetic" undercuts that. It doesn't, if the underlying narrative is genuine. A talking duck selling insurance is synthetic. It is also one of the most effective brand characters in advertising history because the story it told was true: Aflac pays. The medium was artificial. The message was not.
The same logic applies to AI-assisted creative. Generative AI used to tell an authentic brand story is not deception. It is craft. A label saying "this image was AI-assisted" does not diminish the campaign if the campaign is honest about what it is selling.
The brands that will struggle with AI labeling are the ones using AI to fabricate claims, not to produce creative. Transparency requirements are a problem for misinformation, not for good advertising.
This is the argument worth making internally when AI content labeling comes up in brand reviews or client conversations. The question is not whether the tools are synthetic. The question is whether the story is true.
What Brands Should Actually Do Now
The EU AI Act applies extraterritorially on GDPR logic: any provider or deployer whose AI output reaches EU users is in scope, regardless of where the brand is headquartered. For US brands with any EU audience, that is worth taking seriously.
A few practical steps:
- Audit your creative tools. Know which tools in your stack generate synthetic content and whether those providers have signed the EU Code of Practice. OpenAI, Google, Meta, Microsoft, Anthropic, and Mistral are all signed. Tools without a conformity date are a procurement risk.
- Flag deepfake-adjacent formats. Any campaign that uses AI to depict real people, real places, or real events in ways that could be mistaken as authentic needs a visible disclosure. This is the narrowest but most consequential part of the regulation.
- Document human editorial review. For AI-assisted text that touches public-interest topics, maintaining a record of human review provides the editorial exemption under Article 50(4). This is already good practice; now it has a compliance rationale.
- Get ahead of the conversation. Clients will ask about this. Having a clear, accurate answer, specifically that the marking duty sits with the tool providers and that most brand advertising falls outside the visible-label requirement, is more useful than a vague reassurance.
The regulation is new. Enforcement is live as of August 2, 2026. The brands and agencies that understand the mechanics will spend less time in reactive mode when the questions start coming.
