Making AI videos for clients: are you an AI Act deployer?

If you make AI videos for clients as paid or freelance work, the Commission's FAQ points to you being a deployer. It says personal, non-professional use is excluded, but an activity that earns an economic benefit on a regular basis, or is part of a business, trade, occupation or freelance activity, is professional. This is a summary of that FAQ, not legal advice.

The wording below is from the Commission's Article 50 FAQ, read 2026-10-01. Sume details are from Authentication.

Where does the FAQ draw the personal versus professional line?

It gives an example: a person generating deepfakes and sharing them on social media in a personal capacity is a personal activity, outside the AI Act. The same person is a deployer when the use is professional.

What does being a deployer ask of me for video?

The FAQ says deployers must clearly label deepfakes, and that the disclosure must reach a person upon first exposure at the latest, in a clear and distinguishable manner. Machine-readable marking added by the provider does not replace that visible disclosure. Whether a given clip is a deepfake is its own test; see the three criteria.

How does one Sume workspace map to a one-person studio?

Sume resolves workspace, owner and API key metadata from the key, and the docs tell you not to put workspace_id, owner_user_id or user_id in request bodies. A solo freelancer can run one key per client project and keep job records per project; the evidence log post covers what to keep.

Which Sume tools help me apply a visible disclosure?

Standalone captions burn text onto an existing public video URL. With cues (each with text, start and end) you can burn authored overlay text without speech-to-text. Video trim cuts a range from one workspace clip. These prepare the file; the decision about wording and placement stays with you and your client.