Data handling

This page describes how the YAHO API handles the content you submit. It supplements, and does not replace, the Terms of Service and Privacy Policy.

What is stored

You submit	The API keeps
A video (URL, file, batch, sheet or MRSS item)	The media file, the detected scenes with their analysis (captions, translated transcripts, on-screen text, labels, embeddings), keyframes and scene clips, and the video-level summary and labels.
An ad creative	The media file, its analysis, and every screening result produced for it.
A live stream	The manifest URL and, for each ad break, the captured segments' analysis (bursts). The segments themselves are not kept as a library.
A custom summary request	Your instructions and the generated summary.

Everything is scoped to the organization that owns the API key. Another organization cannot read, list or search it.

Names and speech in your content

Scene analysis records the names of people that are written on screen or spoken, together with a translated transcript and visible text, so you can search and filter by them. The API does not perform facial recognition or any other biometric identification; a person who is not named on screen or in speech is described only generically (for example people: [], human_activity: ["presenting"]).

You are responsible for having the rights to process the content you submit, including any personal data it contains. If you need the analysis of a video removed, delete the video (below).

Signed media URLs expire

keyframe_url, clip_url, thumbnail_url and the GET /api/v1/ads/{ad_id}/stream redirect target are presigned URLs valid for about one hour from the moment they are issued. Render them straight away and fetch the scene or ad again when you need a fresh link. Do not store them, email them or embed them in pages that outlive the hour.

Deleting data

Data	How to delete
A video and everything derived from it (scenes, clips, keyframes, embeddings, summaries, custom summaries)	From the dashboard's video library. The video disappears from listings and search for your organization.
An ad creative and its analysis	From the dashboard's Ads tab. Screening results for that creative remain in the Ad Blocking history until deleted there.
A live stream and all of its bursts	DELETE /api/v1/live/streams/{stream_id}. The cascade is immediate and permanent.
An MRSS feed	DELETE /api/v1/mrss-feeds/{feed_id}. Videos already indexed from the feed stay in your library until you delete them individually.
An API key	From the dashboard's API Keys tab. Revoking a key does not delete the data it created.

Deletion endpoints for videos and ads on the public API are not available yet; use the dashboard. Content is retained until you delete it or your organization's account is closed.

AI-generated output

Captions, transcripts, summaries, labels, relevance scores, match reasons and brand-safety verdicts are produced by machine-learning models. They can be wrong, incomplete or inconsistent between runs, and the vocabularies and models behind them are improved over time. Treat them as signals for your own systems and reviewers:

• A brand-safety block is a recommendation to hold the creative for review, not a determination that it violates any law or policy. An allow does not certify that it is compliant.
• Do not surface a screening_failed: true verdict to end users as a content judgement. See Brand safety.
• Labels follow the IAB and GARM vocabularies as implemented by Pyler and are not endorsed by those bodies.

The final decision to serve, block or publish content is yours.