AI Transparency Statement

Draft for counsel review. This document is not in force yet. Highlighted brackets are facts the company still has to supply. The plain-language summaries it replaces stay in force until this version is published without this notice.

How Yoob tells people they are talking to AI, and what we do and do not mark.

In the Companion

The conversation screen shows a persistent label saying the character is an AI, and long sessions repeat it. Characters are instructed never to claim to be human, and to say they are an AI when asked. See the Companion safety protocol.

In apps built with the SDK

Our Acceptable Use Policy requires every app to make it clear to users that they are talking to software, and to never present a Yoob character as a real person. Under the EU AI Act (Article 50) the provider of the app is responsible for that disclosure to its own users; the SDK leaves the screen to the app, so the app chooses how to show it.

Marking of output

Rendered frames are drawn on the user’s device and are never sent to us, so we do not watermark video frames. Each customer’s model weights carry a per-licensee fingerprint that identifies the source of a leaked model; it does not mark the frames themselves. [OWNER: state whether Yoob voice audio carries an audio watermark, and whether C2PA metadata is planned]

Real people

We do not build a character of a real person without that person’s documented, informed consent, including public figures, at any price.

Last updated [PUBLICATION DATE]. Questions about this document: legal@yoob.com.