Why your AI content needs a label

A few weeks ago, I changed my profile picture. I generated it with Gemini and added a line under my About Me section that it was AI-made. On 2 August 2026, that small note stops being a courtesy and becomes the law.

The EU AI Act applies to any provider or deployer of AI systems operating in or serving the EU market, regardless of where they are established. Article 50 sets out the transparency obligations that directly touch marketing work. So let me share what changed, what the rule says, who it reaches — including brands outside the EU — and what you and your team should do before August.

What changed

For years, AI-generated content circulated without a tag. A synthetic image here, an AI-drafted post there. You had no way to tell. From 2 August 2026, if you use generative AI to produce or alter content that reaches the public, disclosure becomes the rule. 75% of marketers have already adopted AI, and creating content is consistently the leading use case.

There is action for two actors. Providers (the model builders) carry the machine-readable marking duty: every AI output must be detectable at the source. Deployers (the brands and marketing teams) carry the disclosure duty: you must tell the audience when what they are seeing was made or altered by AI. Since we count as deployers, this means three things for marketers:

1. Deep fake images, audio, and video must be disclosed.
2. AI-generated text published on matters of public interest must be disclosed.
3. A chatbot has to say it is a chatbot.

The deadlines and the cost

The deployer disclosure duty applies from 2 August 2026. Breaching Article 50 transparency obligations results in fines of up to 15 million EUR or 3% of worldwide annual turnover, whichever is higher. For SMEs and start-ups, these fines are capped at whichever is lower.

There is one critical exception. Public-interest text that goes through genuine human editorial review can be exempt. “Genuine” means a documented process where a natural or legal person holds editorial responsibility for the publication. This does not apply to images. Use of AI for personal, non-professional activity also falls under exemption.

Does this reach you outside the EU?

Yes. Article 2 applies the Act to providers and deployers based outside the EU where the output produced by the AI system is used in the EU. If your content reaches an EU audience, you are in scope on 2 August 2026. There is also an audience reason, not just a legal one: in a Gartner survey of 1,539 US consumers, 68% said they frequently wonder whether what they see is even real.

Four moves before August

1. Decide your compliance path: code or own measures. You can sign the EU Code of Practice by 22 July 2026 or build your own measures. Signing signals intent to follow common standards for marking and labelling. Non-signatories must be prepared to build and defend a custom compliance system for national authority review.
2. Map your AI-generated content by type. Inventory every channel and asset type where generative AI produces or alters consumer-facing content — across social, email, paid, web, and chat.
3. Develop AI disclosure principles. A clear and distinguishable disclosure of the content’s artificial origin, conforming to accessibility requirements, at the latest at the time of first exposure. The EU has published icons for voluntary use:

4. Operationalise the editorial review policy. Establish internal policies that identify a named person or function with editorial responsibility for AI-assisted publications. Document the process to demonstrate genuine human oversight before publication. The exemption is only as strong as the evidence behind it: define who holds responsibility, what the review entails, and how it is recorded.

← All insights