AI labeling requirement from August 2026: What’s really coming your way

From August 2nd, 2026, a new rule applies across the EU: AI-generated content must be labeled in certain cases. This is based on Article 50 of the EU AI Act. In line with this, the European Commission published a Code of Practice in early June, showing how this labeling can work in practice.

“Transparency builds trust.”
— Henna Virkkunen, Executive Vice-President for Tech Sovereignty, Security and Democracy, Source: European Commission

The goal is more transparency: people should be able to recognize when they’re talking to an AI or looking at AI-generated content.

Who does this affect?

• Chatbots & voicebots: Users must be able to recognize that they’re talking to an AI – unless it’s obvious anyway.
• Texts, images, videos: As soon as AI content is published and could give the impression that a human created it alone, it needs a label.
• Deepfakes: Deceptively realistic AI images or videos of real people, places, or events must be clearly marked as artificially generated.
• Emotion recognition & biometric categorization: Affected individuals must be informed when a system analyzes, for example, facial expressions or voice.

When do you need to label content?

The good news: for most marketing texts and social media posts, nothing changes. The requirement mainly applies to topics of public interest such as politics, health, or the economy – but not to classic advertising. The following table shows typical cases from everyday content work:e economy – but not to classic advertising. The following table shows typical cases from everyday content work:

The labeling requirement starting in August 2026 is no reason for concern, but it’s a good opportunity to take a conscious look at your own use of AI. For most marketing content, nothing changes. In the end, the key question always is: could someone mistake the content for purely human-made or genuine? Anyone who keeps that in mind is on the safe side.