EN FI
Insights

AI-generated content labelling becomes mandatory in August

Close-up of hands typing on a laptop with a content-editor interface visible on the screen, placed on a wooden table with a blurred coffee cup in the foreground.

Written by

Riikka Flinkman

Details

3–4 minutes

AI has become a standard part of every marketer’s and communicator’s toolkit. Images, text, videos and audio are now produced so fluently that the end result is often indistinguishable from human-made content. This rapid development has set the legislative wheels in motion: the EU’s Artificial Intelligence Act introduces new requirements from 2 August 2026 onwards, including rules on the identification and labelling of AI-generated content.

What is it about?

The AI Act (EU 2024/1689) is a European Union regulation that establishes common rules for the use and development of AI systems. Its aim is to promote trustworthy and safe AI across Europe. It applies to both developers of AI systems and all businesses that deploy AI or procure AI solutions from third-party providers — which almost certainly includes you, if your organisation uses any AI tools at all.

2 August 2026 marks a practical turning point: that is when wide-ranging transparency and labelling obligations come into force, designed to ensure people always know when they are interacting with AI.

Article 50 of the Act requires AI-generated content to be labelled. A key distinction: AI-modified content refers to material created by a human and then optimised or edited by AI, while AI-generated content refers to material created mainly or entirely by AI. Particularly strict requirements apply to deepfakes — content that resembles a real person or event.

What does the obligation mean in practice?

People have the right to know when they are engaging with AI-generated content. The obligation does not apply only to AI companies and their products — it also covers businesses that use ready-made AI tools. In other words, most likely you.

What matters is the human’s role in the final output:

  • Labelling is generally not required if you wrote or created the content yourself and are only using AI as an aid — for proofreading, summarising, or optimising.
  • Labelling is required if the content is wholly or mainly AI-generated, whether that’s an image, video, or text.
  • Visible labelling is mandatory for deepfake-type content that resembles a real person or event.

How you label depends on the channel and content type: images can carry a technical watermark or a text notice alongside them; written content can include a note at the beginning or end of the article. The key is that the user understands AI’s role.

Checklist for your organisation

  • Does your website use an AI-powered chatbot or customer service automation? –> Inform users clearly.
  • Are you producing images or videos with AI? –> Label them – with text or a technical identifier.
  • Are you producing written content entirely with AI? –> Labelling recommended.
  • Are you using AI only for proofreading or optimisation? –> Labelling generally not required.
  • Are you producing deepfake-type content? –> Visible labelling is mandatory.
  • Does your organisation have an internal policy on what counts as AI-generated content? –> If not, create one now.

What should you do now?

The boundaries are not always clear-cut. The AI Act is high-level, and more detailed guidance is still being developed — the EU AI Office is currently working on a Code of Practice covering the labelling of AI-generated content and the detection of deepfakes. The most important thing is to have a clear picture of where and how AI is being used in your organisation, and to have internal labelling policies in place.

We’re following developments with great interest and will keep you updated! If AI is part of your website or digital services, we can help you assess what changes may be needed. Get in touch!

Sources: