AI Act: labelling AI-generated content becomes the norm IA

Since 2 August 2026, Article 50 of the European AI Act has introduced new transparency obligations around AI-generated or AI-manipulated content. The aim: to make it easier for everyone to tell when they are interacting with an AI, or when a piece of content has been created or modified by one.

Inevitably, for brands, agencies and creative teams building more and more AI into their production, this changes a few of the rules. Here are some examples:

example 1

Some AI-generated content will have to carry a label

Synthetic text, images, video and audio are all covered. AI providers will have to embed markers that make it possible to identify content's artificial origin. Exceptions do exist, depending in particular on the use case and the context.

example 2

Transparency is stepped up around deepfakes and matters of public interest

Deepfakes will have to be clearly flagged. The same goes for certain AI-generated or AI-manipulated content on matters of public interest, when it is published without human oversight or editorial review.

For brands and agencies, one thing therefore becomes essential: keeping a human eye on what gets published.

example 3

Transparency is also a question of trust

Because beyond compliance, a real question arises for brands: how do you talk about your use of AI without devaluing what it produces?

The challenge may no longer be hiding that AI was involved, but finding the right way to signal it without turning every piece of content into a regulatory label. The EU is planning tools and icons to help identify AI-generated content. And on penalties, failing to meet certain obligations can cost a company up to €15 million or 3% of total worldwide annual turnover.