AI-generated content and the EU AI Act: What every content creator needs to know?
How to manage your digital assets more efficiently with a DAM system?
How is AI changing brand asset management?
Try ImageBank X free for 14 days!
Artificial intelligence has significantly accelerated the production of images, videos and other digital content in recent years. Today, an increasing share of content is created with the help of AI. At the same time, this has introduced a new challenge: how can people tell whether content is authentic or has been generated or modified using AI?
The EU Artificial Intelligence Act, or AI Act, addresses this issue through transparency requirements. The obligations concerning AI-generated content under Article 50 began to apply on 2 August 2026.
In this article, we explain when AI-generated content must be labelled under the AI Act and how organisations can ensure that information about the use of AI stays with an asset, even when the person who created it is not the person who publishes it.
AI Act: when must the use of AI be disclosed?
Under the transparency requirements of the EU AI Act, organisations publishing content must disclose the use of AI when the content is:
- A deepfake, meaning an image, video or audio recording generated or modified using AI that resembles an existing person, object, place, entity or event and may appear authentic to the viewer.
- Text published to inform the public on matters of public interest that has been generated or modified using AI without human review and for which no natural or legal person holds editorial responsibility.
For text, a visible disclosure is not required if the content has been reviewed by a person and an organisation assumes editorial responsibility for its publication. This means that not all AI-generated content needs to be labelled in the same way. For example, AI-generated subtitles would not generally be subject to the same disclosure requirements as content that could appear authentic and mislead the viewer.
A useful rule is to ask: could someone mistake this content for something authentic? If the answer is yes, an AI label should be added.
Organisations can, of course, disclose the use of AI more broadly than the law requires. A clear and consistent approach makes everyday work easier and improves transparency across all published content.
How to label AI-generated content: what labels does the EU recommend?
The EU has published three icons for labelling AI-generated content:
- Basic icon: AI has been used to generate or modify the content.
- Fully AI-Generated: The entire content has been generated using AI without any human-created elements or editorial review. Writing a prompt alone does not count as a human-created element.
- Partially AI-Modified: Human-created content has been partially modified using AI.
The EU recommends adding a clear written explanation alongside the icon. The label must be visible no later than the first time a person encounters the content, and it should remain visible when the content is downloaded or shared.
The biggest practical challenge arises before publication
Adding a label to the final publication is rarely the most difficult part. The more challenging question is how the publisher can reliably determine whether AI has been used.
Consider a typical situation. A graphic designer creates a campaign image using AI and saves it in a shared folder. The marketing team then publishes the image across the organisation’s different channels. If information about the use of AI has not been recorded alongside the asset, the publisher must investigate the matter separately or make an assumption.
The same asset may be used on a website, social media, in a newsletter, a dealer portal and printed materials. A note in an email or project management tool does not stay with the file as it moves between people and channels. Organisations need a single piece of information that travels with the asset throughout its entire lifecycle.
How to manage AI labels with a DAM system
When an organisation centralises its assets in a DAM system such as ImageBank X, it can add a dedicated metadata field labelled “AI used in this asset”. This can be a simple checkbox completed by the person creating or uploading the asset.
In practice, the workflow looks like this:
- The content creator uploads the asset to ImageBank X.
- They select “AI used in this asset” in the metadata.
- If necessary, they specify whether the content is fully AI-generated or partially modified using AI.
- The information remains attached to the asset and is visible to everyone authorised to use it.
- Before downloading or publishing the asset, the publisher checks the information and adds the appropriate label for the channel.
The publisher no longer needs to search through messages, folders or filenames for this information. They can immediately see which assets involve AI and what needs to be checked before publication. The content creator records the information once, and the publisher can find it whenever it is needed.
Store your AI labels in your DAM system
The EU’s AI labelling icons should also be stored in your organisation’s DAM system. Add clear usage instructions alongside the icons and, if necessary, limit access to the teams responsible for preparing and publishing content.
When approved icons and instructions are stored in the same place as the assets being published, each team does not need to download them separately or create its own versions. The correct file formats, colours and accompanying explanations are available to everyone.
One place. One set of instructions. One consistent approach for every publisher.
Watch our webinar on AI and brand management
AI labels are one part of responsible brand management. Organisations must also manage usage rights, maintain brand consistency and ensure that approved assets reach the correct channels.
In our webinar, Brand Management in the Era of AI: Opportunities and Risks, we explore how AI is changing brand asset management. You will gain practical insights into how AI and DAM can make assets faster to organise, find and reuse while allowing your organisation to retain control. Get the free webinar recording delivered to your inbox.
Save a full working day every week!
Clear AI labels and well-managed metadata do more than help keep publications compliant. They also make everyday work faster and easier for the entire team. ImageBank X users save their teams up to eight hours every week because they no longer need to spend as much time searching for, checking or verifying assets.
