As of August 2, 2026, new transparency and labelling obligations for certain AI-generated or AI-manipulated content will apply under the EU Artificial Intelligence Act (“AI Act”).
Here is the most important information for directors at a glance.
A compact summary for upcoming commercial projects, personal/spec projects, and the handling of existing works on your showreels.
The European Union’s ‘AI Act’ (Regulation (EU) 2024/1689) mandates, among other things, a labelling obligation for certain image, audio, and video content created or manipulated with the help of AI that qualifies as so-called ‘deepfakes’.
Crucially, it is not merely the use of AI itself that matters, but the overall impression the finished work creates on the audience.
Labelling is particularly required when AI content could appear to viewers as authentic footage (so-called ‘deepfakes’).
For example, in the case of:
✔ Photorealistic AI-generated persons
✔ Realistically sounding synthetic or cloned voices
✔ Realistically appearing locations, environments, objects, or events
✔ AI-generated scenes that appear documentary-style and/or authentic
Rule of thumb:
The more realistic an AI content item appears, and the more likely the audience could mistake it for authentic footage, the more carefully a labelling obligation should be assessed.
The rules apply directly as of August 2, 2026.
Future Commercial Projects
According to the current status and the view of the legal counsel of the German Producers Alliance Advertising Section (Produktionsallianz Sektion Werbung), the ultimate responsibility for the final decision regarding potential labelling typically lies with the client (advertiser), as they decide on the publication and dissemination of the content.
The production company, in turn, should implement the labelling agreed upon with the client in the master versions intended for release. This also applies to Director’s Cuts.
For us as directors, we nevertheless recommend:
✔ Make your use of generative AI transparent to the production company and/or your client at an early stage. The final decision regarding the use of AI in the commercial should always rest with the client.
✔ Keep a record of which AI-generated or AI-modified content you personally introduced into the process, and inform the production company thereof in writing.
✔ Address the labelling issue with the production company early on (e.g., during the PPM or, at the latest, at the start of post-production) so they can clarify the matter with the client in a timely manner.
Personal / Spec Projects
For your own personal projects, where you use AI under your own professional responsibility and make the work publicly accessible yourself, you bear full legal responsibility for any required labelling, provided the work meets the criteria outlined above.
You are fully and legally responsible for your own channels. This explicitly applies to your Director’s Cuts and personal projects showcased on your own platforms: Your Website, Vimeo & Social Media.
As soon as you publish films yourself, you should assess whether labelling is required for that specific content. Even if a film was originally created for a client, a re-assessment can be advisable for your own publication. We recommend coordinating closely with the respective production company.
The regulation does not have retroactive effect.
In accordance with the current guidelines of the EU Commission, AI-generated or AI-modified image, audio, or video content created or edited prior to August 2, 2026, does not need to be labelled retroactively. This also applies if it remains publicly accessible on your website or channels after the cutoff date.
Voluntary retroactive labelling is recommended where possible without disproportionate effort. However, there is no legal obligation to retroactively edit older works or remove them from websites and social media profiles. If an older work is re-edited using AI after August 2, 2026, the new version must be assessed separately.
It is not solely relevant where a work was produced. If AI-generated or AI-modified content is specifically published within the EU or made available to an EU audience, the transparency obligations of the EU AI Act apply equally. The responsibilities are distributed as outlined above.
There is currently no binding industry standard. However, the EU Commission has made available optional labelling icons for fully AI-generated or partially AI-generated content. These are available here.
Labelling can be accomplished using an appropriate EU icon and/or a correspondingly clear text disclosure. Depending on the content, it may be useful to supplement the icon with a short explanation, for example, if only a voice or a single scene is affected. Text disclosures may include, for example: • “AI-generated” • “This scene contains AI-generated content.” • “Voice AI-generated.” • “Individual scenes were created or edited using AI.” (or similar)
The disclosure must be clear and distinguishable, appearing at the latest upon the first perception of the content. For moving image content, legible labelling within the frame or an immediately associated platform label is generally the safest solution. It should remain visible for a sufficient duration; there is no fixed minimum time requirement.
Disclosures hidden in captions after “See More” or placed in obscure locations are generally insufficient.
A violation of the labelling obligation on your own channels can carry various legal consequences:
Account / Content Bans: Platforms such as Vimeo or Instagram may, in accordance with their terms of service, mute, deprioritize, or remove your videos.
Cease-and-Desist Letters (Abmahnungen): Provided that additional requirements under unfair competition law are met, unlabelled AI usage on your own channels can trigger legal warnings and cease-and-desist claims.
Worst-Case Scenario: If you conceal the use of AI from the production company or your direct client, you may face severe claims for damages/recourse due to a breach of your duty of disclosure, should they later be fined as a result of your commercial.
Not every use of AI falls under the new transparency obligations. As it currently stands, this generally does not apply to the use of AI as a purely internal production tool, such as for:
Treatment texts
Moodboards
Pre-visualizations (provided these contents are not published themselves or do not become part of the final commercial asset).
The same applies to surreal or manifestly artificial visual worlds that no viewer could reasonably mistake for reality.
Furthermore, AI-supported tools in the field of VFX do not automatically trigger a labelling obligation, provided they are purely technical/craft-based in nature and merely optimize existing footage. These include:
Digital clean-up / retouching
Color grading
Crowd extensions
Weather effects
(Note: Substantial crowd extensions, weather effects, and major background alterations should nevertheless be evaluated on a case-by-case basis as a precaution.)
The line in VFX work may be crossed when AI creates or substantially alters persons, voices, locations, objects, or events to appear deceptively authentic, thereby conveying a false impression of reality.
Before every publication, a quick check is recommended:
✔ Does the finished work contain realistically appearing AI content (visual & audio)?
✔ Could viewers mistakenly mistake this content for authentic footage?
✔ Have I transparently informed the production company about my AI usage in the final work?
✔ Does the publication on my channels contain the required disclosure, clearly recognizable upon first perception?
Disclaimer
The interpretation and practical implementation of the new EU regulations continue to evolve. The guidelines published by the EU Commission, the Code of Practice, and the new EU icons provide initial guidance. Industry standards as well as initial administrative and judicial decisions are expected to provide further clarity. This information sheet serves orientation purposes and does not replace legal advice in individual cases.
We ask for your understanding that DRCT cannot provide individual legal counsel. Whether and how labelling is required depends on the specific content and context of use. In cases of doubt, legal counsel should be sought. General questions from members will be bundled and addressed in future communications where possible.