Effective August 2, 2026, the key transparency requirements of the European AI Act will take effect. Under these requirements, companies will be required in certain cases to disclose whether content has been generated by Artificial intelligence were generated or manipulated. Learn when AI labeling is required and what companies need to keep in mind.
The Key Requirements of Transparency Obligations
The AI Regulation does not introduce a general labeling requirement for all AI-generated text, images, or videos. The transparency requirements under Article 50 of the AI Regulation apply in particular to deepfakes and to certain AI-generated content that is published to inform the public about matters of public interest.
Effective August 2, 2026, the following requirements will apply in particular:
- Users must be able to tell when they are interacting with an AI system.
- Providers of generative AI systems must provide technical mechanisms for labeling synthetic content.
- Deepfakes must be disclosed.
- The operator is subject to additional transparency requirements for certain AI-generated content.
This presents companies with two key challenges: First, they must identify when transparency requirements apply; second, they must be able to determine where AI is actually being used within the company.
Provider and Operator Obligations: An Important Difference
For practical implementation, the distinction between providers and operators of AI systems is crucial.
Providers of generative AI systems, such as OpenAI, Google, or Anthropic, must provide technical mechanisms that can be used to identify artificially generated content.
Companies (operators) that use such systems, however, are subject to transparency requirements for specific use cases. Of particular relevance here are deepfakes and certain AI-generated content of a public informational nature.
Case Study: Mandatory Labeling of AI-Generated Texts
ChatGPT as an Assistant Tool
An employee uses ChatGPT to write a professional article. The AI provides him with suggested outlines and draft text. The article is then fact-checked, revised, and approved by the editorial team.
In this case, there is generally no separate disclosure requirement. The individual retains control over the content and assumes editorial responsibility.
Automatically Published Information
The situation may be different when AI-generated content is published automatically and serves to inform the public.
For certain AI-generated content intended to inform the public about matters of public interest, Article 50 of the AI Regulation establishes transparency requirements. At the same time, the regulation contains an important exception: If human verification or editorial oversight takes place and a person or organization assumes responsibility for the content, the disclosure requirement generally does not apply.
Practical Tip
Companies should document,
- if content was generated entirely by AI,
- whether a human review took place,
- who is responsible for editorial content.
This information may be relevant in the context of compliance or regulatory audits.
Case Study: Labeling Requirements for AI-Generated Images
AI-generated illustrations
A company uses Midjourney to create images of employees or office spaces that do not actually exist.
However, the mere fact that an image was generated by AI does not automatically trigger a labeling requirement. The issue of transparency becomes particularly relevant when the image appears realistic and users might assume it is an authentic photograph.
In such cases, an assessment should be made as to whether transparency obligations under Article 50 of the AI Regulation are triggered.
AI-powered image editing
An existing product photo is cropped, optimized, or adjusted for color and lighting using AI.
No new realistic scenario is created here. The AI merely assists in editing an existing image. Transparency requirements are generally not triggered in such situations.
Practical Tip
The more closely a depiction simulates real people, places, or events, the more careful the legal assessment should be.
Case Study: AI-Generated Videos and Deepfakes
What is a deepfake?
Article 3, No. 60 of the AI Regulation defines deepfakes as image, audio, or video content generated or manipulated using AI that resembles real people, objects, places, or events and can thereby convey an impression of authenticity.
Special transparency requirements apply to such content.
AI Avatar as a Corporate Spokesperson
A company uses an AI-generated avatar as a corporate spokesperson for product presentations or customer information.
If users cannot tell that a representation is artificially generated, appropriate transparency measures must be taken.
Deepfake of a CEO
As part of a marketing campaign, a video is released that appears to feature the company's CEO speaking. In fact, both the voice and the image were generated entirely by AI.
In this case, the artificial nature of the content must be clearly disclosed.
Practical Tip
When it comes to deepfakes, the Transparency be made directly within the content. In addition, it is recommended to Documentation the AI systems used and the measures taken.
How should AI be labeled?
The AI Regulation does not prescribe any specific format or content for labeling. However, companies must ensure that users can clearly identify content that has been artificially generated or manipulated.
It is crucial that the disclosure be clear, understandable, and not concealed.
The following phrases can serve as a guide:
Labeling AI-Generated Images
Possible clues include, for example:
- „This image was created using artificial intelligence.”
- „AI-generated illustration.”
- „This image was generated with the help of an AI system.”
The note should appear directly next to the image or in the image description.
Labeling AI-Generated Videos
For videos, it is recommended to include a label directly in the video and also in the video description.
- „This video contains AI-generated content.”
- „The person depicted was generated using artificial intelligence.”
- „The voice and image were generated by AI.”
Labeling Deepfakes
In the case of deepfakes, it should be made particularly clear that the content has been artificially generated or manipulated.
- „This video was created using AI and does not contain any actual footage.”
- „The person depicted did not actually make the statements shown.”
Labeling AI-Generated Text
If there is a disclosure requirement, a notice might read, for example:
- „This content was created using artificial intelligence.”
- „The following post was generated in whole or in part by an AI system.”
Note: Currently, there are no binding standard guidelines for labeling AI content, either at the European level or from national regulatory authorities. Companies should develop appropriate Guidelines the European Commission and the relevant market surveillance authorities will therefore be closely monitoring the situation.
Labeling Requirements Begin with AI Governance
For many companies, the real challenge lies in identifying the affected use cases.
Those who do not know which AI systems are being used can hardly meet transparency requirements reliably.
That is why companies should create an AI inventory and, in particular, document the following:
- AI systems used,
- Intended uses,
- responsible Departments,
- generated content,
- human control mechanisms,
- Relevant AI Act obligations.
The AI labeling requirement is therefore not an isolated marketing issue, but rather part of a comprehensive AI governance and compliance strategy.
Penalties for Violations of the AI Labeling Requirement
The transparency requirements are legally binding.
The AI Regulation provides for substantial fines for violations of transparency requirements. The specific amount depends on the applicable penalty provision and the circumstances of the individual case.
Depending on Infringement Fines of up to 15 million euros or up to 3 % of global annual revenue may be imposed.
Do you have questions about implementing the AI labeling requirement?
The requirements of the AI Regulation go far beyond the labeling of individual pieces of content. From establishing AI governance and creating an AI inventory to implementing transparency and documentation requirements, companies must meet numerous new requirements.
Our experts will assist you in implementing the requirements of the AI Regulation in a legally compliant manner and integrating them into your existing compliance and governance structures.
Please contact us for a no-obligation consultation.
German 
English 
Italian 
French