Images exported from Qamera have carried a signed record of their origin since early August. Video did not carry anything at all — it was the one format leaving the platform with nothing to say about where it came from. That is now fixed.
Every video generated from today onwards leaves with a signed manifest embedded in the file itself. It states that the material was produced by an AI model, names the platform that produced it, and carries a timestamp. The signature is bound to the exact bytes of the file, so any later alteration of the video is detectable rather than invisible.
Where the video came from before us is preserved too. Some of the models we use sign their own output. Rather than replacing that signature, we keep it: the finished file carries the model provider's record as well as ours, so the chain stays readable end to end instead of stopping at the last step.
Your videos are unchanged as videos. The record is metadata alongside the picture, not a change to it — the frames are identical, and the file grows by about a hundredth of a percent. Nothing about playing, downloading or publishing your videos works differently.
Why this matters. From December 2026 European rules require AI-generated material to be marked in a way machines can read, and the same information increasingly determines how platforms treat what you publish. This is that marking, applied to a format where it was missing — with no step added to your workflow and no cost to your credits.
Videos generated before today are unaffected and keep working exactly as they did.