An AI-generated shot can pass through several people before anyone publishes it. The information about how it was made needs to travel with it. That is the practical production issue behind the EU’s current work on AI-content transparency.
The European Commission’s task-force update, dated October 28, 2026, records an interoperability working-group meeting on October 23, with the next scheduled for November 5. The group is working toward making marking and detection systems work together, so people do not have to try several detectors independently.
For editors and agencies, the useful question is what happens between generating a clip and delivering the final video.
What is changing, and what is already in force?
The Commission describes two signatory task forces established on September 30, 2026: one for marking and detection, and one for labelling. Work includes disclosure practices and development of the EU label. Further groups on an interactive information layer and an audio disclosure signal are planned for January 2027.
These are implementation activities, not a newly adopted law or a finished universal detection system.
The underlying EU AI Act transparency obligations became applicable on August 2, 2026. The Commission’s Code of Practice overview distinguishes providers’ machine-readable marking and detection duties from deployers’ disclosure duties for deepfakes and certain text publications. Signing the code is voluntary; the applicable legal obligations are not.
That distinction matters in production. A visible caption and a technical marking perform different functions. Deciding what disclosure a particular video needs depends on its content and the applicable rules, not simply on whether AI was used somewhere in the edit.
Keep the shot’s origin with the edit
Our practical recommendation is to make the handover specific enough that the next editor does not have to reconstruct the production history.
For a campaign combining filmed product footage with generated scenery, keep a small shot record alongside the approved cut:
- The source filename and where the shot appears in the delivered version.
- Whether the image, voice or another element was generated or materially altered.
- The original generation output and any available provenance information.
- The approved disclosure wording and where the publisher intends to show it.
Use descriptions that distinguish the elements. “Generated background behind filmed product” is more informative than labelling the entire folder “AI.” When a shot is replaced, update its record too.
This is a production practice, not a substitute for deciding the relevant legal obligations. It helps the person making that decision understand the actual video.
Review the delivered version, not only the working timeline
A disclosure approved for a wide version may be hard to read in a vertical crop. A shorter edit may omit the closing slate where the explanation appeared. Include the disclosure in the review of each deliverable, alongside captions, framing and sound.
On Tellers, you can prompt our agent to add approved disclosure text and adjust its placement on the editable timeline. For example: “Add this exact disclosure to the vertical version, keep it clear of the captions, and show me the result before export.” Check the exported file and the publishing platform’s own disclosure controls as separate steps.
Our AI Act overview covers the broader context. The immediate editorial habit is simple: hand over the video and the information needed to describe it honestly.
Frequently asked questions
Is a visible AI label the same as machine-readable provenance?
No. A visible label explains something to the viewer. Machine-readable marking carries information that software can read or detect. Keep the original generation files and any available provenance information alongside the edit; adding a caption does not replace that information.
What should I hand over with an AI-generated video?
Include the final video, a record of the generated or materially altered shots, their source filenames and positions in the edit, and any approved disclosure wording. Identify whether the image, voice, background or another element was generated so the publisher can review the actual content.
Can I add disclosure text to a video by prompting Tellers?
Yes. Give our agent the exact approved wording and ask it to add the text to the editable timeline. Specify placement and duration, then review each exported format for readability. Adding the text is an editing step; it does not determine which disclosure obligations apply to the video.