AI-generated media on this site
Some background videos on this site were generated by an AI model. This page says which, how they are marked, and what that marking means.
What is generated, and what is not
Generated: the decorative background videos in the separating bands between sections. They show no people, no places and no events — they show ink, paper and light.
Not generated: the product film in the header of the home page. It is a screen recording of the real application, recorded in a browser, including a real evaluation by the exam simulator. Nothing in it is staged.
Text, prices and figures on this site are written and measured, not generated. Where the product itself creates AI content for you — a podcast, a summary, flashcards — that is labelled at the content itself.
How a generated video is marked
Three ways, and they do different things:
- Visible, at the band itself: the note that links to this page. For the person looking at it.
-
Machine-readable, in the file: the IPTC field
DigitalSourceTypecarries the valuetrainedAlgorithmicMedia, alongside the provider and the model. Anyone can read it — for example withexiftool. -
Structured, in the page source: the
VideoObjectnode carries the same statement, so that a machine finds it without opening the file.
The model provider additionally embeds an invisible watermark (SynthID) in the picture itself. We never remove it.
Which model
Google Veo, via the Gemini API. The exact model version is written into
each file as CreatorTool, so it stays readable even if we
change models later.
Why we mark at all
Because a moving picture that looks real should say when it is not. We do not claim that the law requires a visible note for decorative background video — that question is genuinely open, and we would rather answer it than sit it out.