GeniusClip beta
Sign in Create my first clipsTry it

Data Collection Consent

Last updated: 2026-09-15

This English version is a translation. If it differs from the French version, the French version applies.

1. What is this document about?

GeniusClip is working to improve its editorial scoring, hooks and titles, eventually with proprietary models trained on real usage data from the Service. This document describes the framework for this voluntary collection.

This collection is entirely optional and revocable. It is off by default. You can turn it on or off at any time, with no loss of features and no price penalty.

This document explains precisely what is covered if you turn collection on, and what is not.

2. What is NOT covered

The following operations work regardless of your choice about collection. They are needed to provide the Service and are covered by the Terms and the Privacy Policy.

  • processing your Source Videos to generate your own Clips;
  • storing your videos, transcripts and clips for your own use;
  • automated processing of your audio to produce the transcript;
  • processing the transcript to generate hooks, titles and scores.

In other words: your decision about this consent does not affect how GeniusClip works for your own use. It only concerns the internal reuse of derived data to prepare a future model.

3. What is covered if you turn collection on

If, and only if, you explicitly turn on the option in Paramètres → Confidentialité → Contribuer aux modèles futurs (Settings → Privacy → Contribute to future models), the following items are copied into an internal dataset:

  • your feedback on clips: clip kept or rejected, hook chosen, reason given;
  • your editorial edits: the titles, hooks, descriptions and hashtags you rewrite;
  • reported performance, if and only if you import a clip's statistics yourself (views, completion rate, shares).

Videos, their audio and full transcripts are never copied into it.

4. What it is used for

The collected dataset is used only for:

  • preparing a GeniusClip proprietary model trained on real usage data;
  • improving scoring (Genius Score) by comparing our predictions with observed reality;
  • optimizing hooks and titles through a detailed analysis of what works in different contexts (business goal, platform, duration).

5. What it is NOT used for

Explicit commitment: the collected dataset will never be:

  • sold to a third party;
  • shared with an advertiser or an ad network;
  • used to train a third party's model;
  • used to target advertising or retargeting;
  • cross-referenced with external databases to profile users.

The dataset stays internal to GeniusClip and to its product improvement purpose. If the business is transferred, the buyer will be bound by the same commitments; any change of purpose would require new explicit consent.

6. Pseudonymization

Before joining the dataset, the collected data is processed as follows:

  • the dataset contains neither your email address nor your Stripe ID;
  • the identifiers of your workspace, projects and clips are replaced by hashes. Your account cannot be traced from the dataset alone, but all your records carry the same workspace hash;
  • links, file names and paths are removed, and strings that look like keys or passwords are masked;
  • limitation: titles, hooks and descriptions may contain identifying information taken from the video (names, places, etc.). The Service does not detect it automatically. If your videos contain sensitive information, we recommend not turning collection on, or withdrawing your consent afterwards (see section 8).

Under the GDPR, this processing is pseudonymization: the data remains personal data and is protected as such. That is why we chose opt-in: to leave users in control of this decision.

7. Why opt-in rather than opt-out?

Three reasons:

  1. Strict GDPR compliance. Article 7 and Recital 32 of the GDPR require consent that is “freely given, specific, informed and unambiguous”, expressed by a “statement or by a clear affirmative action”. A pre-ticked box or an opt-out is not valid. We use pure opt-in.
  2. An ethical choice. We do not want users to contribute to a commercial dataset without being fully informed and in control.
  3. Product strategy. If we ever launch a proprietary model, we want to be able to say publicly that it was trained on data given voluntarily, unlike the opaque practices of some market players.

8. Withdrawing consent

You can withdraw your consent at any time, through the following channels:

  • from your space: Paramètres → Confidentialité → Contribuer aux modèles futurs, by unticking the box. The data already collected is erased immediately and nothing more is collected;
  • by email to privacy@geniusclip.eu: same effect, within 30 days.

Withdrawal is not retroactive for training runs already carried out (a model that has seen a sample cannot formally “forget” it). However, the data concerned is removed from the dataset and will not be used in any later retraining.

Withdrawing consent changes nothing about your access to the Service, its features or its price.

9. Retention

Data collected under this consent is kept:

  • as long as your consent is active, with no fixed time limit;
  • after consent is withdrawn: erased from the dataset immediately;
  • if you close your account (see Data Deletion): deleted within the same 30-day window as the account.

10. Updates

If we decide to widen the purpose of the dataset (for example, training a model sold under another brand, or sharing with a research partner), we will notify you explicitly before the change and ask for new consent. Without a reply, your current consent is considered limited to the purpose described above.

11. Contact

For any question about this consent, withdrawing it, or deleting past data from the dataset: privacy@geniusclip.eu.