Biometric & likeness data consent policy
Voice cloning and the visual AI Twin work with biometric data - your voiceprint and faceprint. This policy explains what we collect, why, how long we keep it, and the consent we require before we ever create it.
The registered legal name and address of the business that operates Repic are not yet published in this document. They will be added here before it leaves draft. Still to be set on this page: the deletion window and the retention ceiling, which wait on a deletion job that does not exist yet.
In plain terms
- If you publish a post that uses your visual Twin as its presenter, the platform keeps that copy under its own terms - deleting your Twin or your Repic account does not remove it (section 11).
- Repic does not yet add its own machine-readable "AI-generated" marking to what it makes, and does not claim to meet the EU AI Act's marking rule; section 12 says exactly what is and is not marked.
- Repic creates a voiceprint (your cloned-voice model) and a faceprint (face-geometry data from your photos) to build your AI Twin.
- We use your biometric data for one purpose only: to create and operate YOUR AI Twin. Never for anything else.
- We ask for your explicit, separate consent first, and you confirm the voice and face are your own - Repic is own-voice and own-likeness only. You must be an adult.
- The in-product consent step is live for voice twins; for visual twins it is still being rolled out (section 5 says exactly where it stands).
- We never sell your biometric data, never profit from it, and never let our providers use it to train their general models.
- Email privacy@repic.site to have your samples and derived biometric data destroyed and any cloned voice revoked - our team carries this out on request while we finish the automatic controls.
Why this policy exists and what it covers
This policy explains how the operator of Repic ("Repic", "we", "us", "our") collects, uses, stores, and destroys biometric and likeness data. It sits alongside our privacy policy and gives the biometric features their own, more detailed treatment, because this data is sensitive and specially protected by law.
It covers the two places where Repic works with biometric data: voice cloning (building an AI version of your voice) and the visual AI Twin, including the face features that keep your generated likeness consistent. If you do not use these features, Repic does not create biometric data about you.
What we mean by biometric and likeness data
We use these terms throughout this policy:
- Voiceprint - the cloned-voice model and voice reference built from a recording of your voice, which lets Repic synthesise speech in your voice.
- Faceprint - face-geometry data derived from your photos: numerical face embeddings, facial landmarks, head pose, and face-parsing region maps that describe the geometry of your face.
- Likeness assets - the generated images and video of your face and body in new poses, expressions, outfits, and scenes, and the try-on composites of you wearing a garment.
- Character-sheet description - a written description of your appearance (for example skin tone, complexion, apparent age range, gender presentation, and distinguishing marks) that helps keep your likeness on-model.
What we collect and how
Voice. When you create a voice twin, you record or upload a voice sample. That sample is processed through Google Cloud Text-to-Speech (Chirp 3 Instant Custom Voice) to build a custom voice reference for your account. The voice reference is held server-side.
Face. When you create a visual AI Twin, you upload reference photos. From those photos Repic builds a character sheet and computes face-consistency identity data - face embeddings, facial landmarks, head pose, and face-parsing - that keeps your generated likeness recognisably you across poses and scenes.
Why we use it
We use your biometric and likeness data for one purpose only: to create and operate your AI Twin - to let it speak in your voice and to keep your generated likeness consistent and recognisably you. We do not use it for any other purpose - not advertising, not analytics about you, not building features for other customers, and not to identify or track you or anyone else.
Your consent
Your explicit, informed consent is the only basis on which we create a voiceprint or faceprint. This consent is separate from, and additional to, your general acceptance of our terms of service, and it applies only to that biometric processing.
For voice twins, the consent step is in the product today: you cannot train a voice without ticking the specific consent box on the recording screen. For visual twins, the equivalent in-product consent step is being rolled out and is not live yet. Until it ships, uploading reference photos to build a visual twin is your confirmation that the face is your own and that you agree to this policy - and you can have that data destroyed at any time by emailing privacy@repic.site. We would rather tell you exactly where this stands than imply a gate that is not there yet.
By giving that consent you confirm that the voice or face is your own. Repic is an own-voice and own-likeness product: you may not build a twin of another person on your account, even with their permission (see section 10 and our acceptable use policy). You confirm that you are an adult. You give the consent freely, and you can withdraw it at any time (see deletion and revocation below). Withdrawing consent does not undo processing that already happened, but it stops any further use and triggers destruction of the data.
Retention and destruction schedule
We keep biometric identifiers only as long as we need them to provide your AI Twin, and then we destroy them. This is the schedule we hold ourselves to:
We retain your biometric identifiers only for as long as needed to provide your AI Twin. We will delete the underlying raw voice sample and reference photos, destroy the derived biometric identifiers (the voiceprint, face embeddings, landmarks, and face-parsing data), and revoke any provider-side cloned voice within [30] days of the earliest of: you deleting the relevant twin, your request to delete, the closure of your account, or the purpose being fulfilled - and in any event no later than [3 years] after your last interaction with the feature.
No sale, no profit, no third-party training
We do not sell, lease, trade, or otherwise profit from your biometric data. We do not disclose it except as described in this policy, and we do not permit the providers who process it on our behalf to use it to train their own general models. Its only role is to power your AI Twin, for you.
Deletion and revocation
You can ask us to delete your biometric data at any time by emailing privacy@repic.site. Deleting the relevant twin in the product is also a request to remove the biometric data behind it.
Acting on a deletion, we remove the underlying voice sample and reference photos, destroy the derived biometric identifiers, and revoke the cloned voice held with our provider so that it can no longer be used. We honour these requests on the schedule above and confirm when it is done.
Deleting your Twin removes it from Repic. It does not remove posts that already carry your likeness on Instagram, Threads or LinkedIn - those are the platform's copies, and you delete them there (section 11).
Deletion of the stored media and revocation of the provider-side voice are carried out by our team on request while we finish building the automatic controls: today, pressing delete in the product removes the twin from your account, and the full destruction described above is completed by us to the schedule in section 6. If you want the underlying files and the cloned voice destroyed now, email privacy@repic.site and we will do it and confirm. We are saying this plainly rather than implying a one-click erase that does more than it currently does.
Who we share it with
Creating and running your AI Twin involves one main processor: Google. Your voice sample is processed by Google Cloud Text-to-Speech to build and run your voice; reference photos and generation run on Google Cloud AI services; and media is stored in Google Cloud Storage. Google acts as our processor, bound by contract to use this data only to provide the service to us, and not for its own purposes.
Your voice reference (the cloning key) is held server-side only and is never sent to your browser. We disclose biometric data only to processors who help us run the feature under contract, or where the law requires it (for example, a valid legal request). We list our processors in the privacy policy and on our sub-processors page.
When you publish a post that uses your Twin, its images go to the platform and account you chose. Your biometric data does not - section 11 explains the difference.
Other people and minors
You may never create a biometric of another person on your own account. If a voice or face is not yours, that person must create their own twin, on their own account, with their own consent - their consent cannot be given by you on their behalf.
Repic is not for anyone under 18. We do not knowingly create a voiceprint or faceprint of a minor, and you must never upload a minor's voice or photos to these features. If we learn that biometric data of a minor has been created, we delete it.
Publishing content that shows your likeness
When you choose "use my twin as the presenter" for a carousel or post - which needs a ready visual Twin and your imagery consent - Repic conditions every image in it on your Twin, so the people in those images look like you. If you then publish that post to Instagram, Threads or LinkedIn, the images carry your likeness to the account you chose, at the time you chose. As with everything Repic publishes, that happens only for a post you scheduled yourself or approved.
What goes to the platform is the post's images. Your biometric data does not: the voice samples and reference photos you gave us, the identifiers derived from them, and the voice-cloning key stay with Repic and the processors in section 9.
Once a post is published, the platform holds its copy under its own terms. Deleting your Twin, disconnecting the account, or deleting your Repic account does not remove posts that were already published - delete those on the platform itself.
How Repic marks AI-generated content
Everything Repic generates - images, video, voice and text - is made by AI, and content that uses your Twin can look and sound like you. The law now asks for that to be marked, so this section says exactly what Repic does and does not do, without claiming more than we have built.
What Repic marks today: nothing of its own. Repic adds no machine-readable marker to what it generates - no Content Credentials (C2PA) manifest, no metadata tag, no watermark - and no visible "AI-generated" label on exported files or published posts. When it publishes for you it does not add a disclosure to your caption or switch on a platform's AI label.
What the underlying models may carry. Google, whose models generate Repic's images and video, says it embeds an invisible SynthID watermark in content those models produce. That is Google's mark, not Repic's: we do not add it, we do not check that a given file carries it, and what Repic does to an image afterwards - laying text over it, cropping it to a platform's size, adjusting colour or contrast - may weaken it. Voice audio from a voice Twin carries no marker that Repic adds.
What the law requires. Article 50 of the EU AI Act has applied since 2 August 2026. It requires providers of systems that generate images, video, audio or text to mark that output in a machine-readable way that shows it was artificially generated, and it requires anyone who publishes a realistic AI-generated image, video or audio resembling a real person to disclose that it is artificial. Under the AI Omnibus agreement, systems already on the market before 2 August 2026 - Repic among them - have until 2 December 2026 to meet the machine-readable marking duty.
Repic has not yet built machine-readable marking of its own, so we do not describe Repic as meeting Article 50's marking requirement. This section will change when that ships, and not before.
Your part. If you publish content made with Repic that shows your likeness - especially where you or your audience are in the European Union - say that it is AI-generated, in the caption or with the platform's own AI label.
Regional notices
Some regions give biometric data specific protection, and those protections apply on top of this policy:
- Illinois (BIPA). We obtain a written release before creating a voiceprint or faceprint, publish this retention and destruction schedule, do not sell or profit from your biometric data, and recognise that Illinois law provides a private right of action.
- Texas (CUBI) and Washington. We obtain your consent before capture, use the data only to provide your AI Twin, and destroy it within the schedule above.
- EEA and UK (GDPR Article 9). Voiceprints and faceprints are special-category data; our lawful basis for processing them is your explicit consent, which you can withdraw at any time.
Because region-specific consent and controls are still being rolled out, the biometric features may be limited or unavailable in some regions until we can offer them in a fully compliant way.
Contact
Questions about this policy, or a request about your biometric data, go to privacy@repic.site. We answer, and we act on requests. You can also reach us through our contact page.