Resources

How to sell your data to AI without losing ownership

Licensing is not a buyout. How to keep your copyright, authorize fiund to handle AI licensing, withdraw from future deals, and get paid per licence.

Published 2026-07-22 · 5 min read

Key takeaways

  1. Selling here means licensing. You grant AI-training rights and keep the copyright.
  2. Every asset carries a signed licence that names AI training explicitly.
  3. fiund exclusively represents active content for AI licensing and can grant it to multiple buyers.
  4. The contributor agreement authorizes ordinary deals; you can withdraw content from future licensing.
  5. Where people are identifiable, voice and likeness consent sits separate from the asset licence.

You have recordings an AI lab would pay to train on. You also do not want to hand them over and never see them again. That fear stops most owners before they start. It rests on a word doing the wrong work here: sell.

On fiund you licence. You keep the copyright and appoint fiund to handle AI licensing for the content you submit. fiund can grant the same recording to multiple buyers, and you can withdraw it from future licensing. None of that is a permanent buyout.

Licence versus buyout

A buyout moves ownership. You are paid once, the copyright changes hands, and the buyer does as it likes. A licence is narrower. You grant specific rights, for specific uses, and keep the title. When the licence covers AI training, it says so in plain words.

Every asset on fiund carries a signed licence that grants AI-training rights explicitly. Not implied. Not buried. If a buyer wants to train a model on your recording, that right is written down and dated. You still own the recording the day after the deal closes.

Scope is the whole point of a licence. It can name the buyer, the model, the purpose, the territory and the term. Anything outside those lines is not granted.

You keep ownership while fiund handles buyer licensing

Ownership does not transfer. You remain the rights holder, keep publishing and monetizing the work normally, and receive a contributor share each time fiund licenses it.

The contributor agreement is the authorization. It appoints fiund exclusively for AI licensing while the content is active, so ordinary buyer transactions do not require another round of emails and signatures. If you want fiund to stop offering an asset for future deals, you can request withdrawal at any time.

One representative, multiple Buyer Licences

Data is non-rival. Two labs can train on the same hour of audio without taking anything from each other. A physical object is sold once. A recording can be licensed many times.

fiund exclusively represents active content for AI licensing, then can grant non-exclusive Buyer Licences to multiple customers. If one buyer wants to block other AI buyers, that requires a separate contributor addendum defining the scope, term, and additional payment.

How you get paid

You are paid per licence. Each grant to each buyer is its own agreement between fiund and that buyer. Because the asset is non-rival, more grants can mean more income from the same material, without giving up ownership or signing another approval for every transaction.

There is a real choice between a recurring arrangement and a one-off. A buyout trades future upside for a single payment. A licence keeps you in the loop. Which fits depends on your material and your appetite, and we cover the trade-off in a separate piece.

Payment follows the contributor agreement and any separate addendum you accept. What matters is that each buyer holds its own grant, and a standard Buyer Licence does not close the door on the next.

Consent is captured at the source

If people are identifiable in your recordings, their voice and likeness are their own. A licence over the recording is not the same as consent from the people in it. Both have to exist.

Where people can be identified, fiund keeps voice and likeness consent separate from the asset licence. That protects the people on the recording, and it protects you as the supplier. Provenance for all of this is available in diligence, so a buyer can check the chain rather than take your word.

Where to start

Begin with what you have and what its rights status is. Clean, consented, well-labelled material moves fastest. Decide which assets you are comfortable appointing fiund to represent for AI licensing, and keep raw versions apart from edited ones.

The vertical guides below go deeper by type: video archives, call-centre audio, research interviews, voice work, and spoken-word collections. Start with the clean shelf and work outward.

Do not wait for the archive to be perfect. A single clean, well-documented asset is enough to begin, and the first listing teaches you what buyers ask for.

NoteA licence grants use. It does not move ownership. fiund can grant the same active asset to more than one Buyer.
Watch outA permanent buyout can look attractive up front. It also ends your say over the material. Read the term before you sign.

Before you list an asset

  • Confirm you own or control the recording, and can prove it.
  • Check whether people are identifiable, and whether you hold their consent.
  • Decide which assets you want fiund to represent for AI licensing.
  • Gather provenance: where it came from, who recorded it, when.
  • Separate raw material from edited versions.

← All resources

Frequently asked questions

Do I lose my copyright when I license to an AI lab?

No. A licence grants training rights for a defined use. You keep the copyright. While the asset is active, fiund handles its AI licensing and can grant it to multiple buyers.

Can I stop future licensing of my data?

Yes. You can request withdrawal at any time. Once it is effective, fiund stops granting new Buyer Licences for that content; licences already granted continue under their written terms.

Is licensing the same as selling my voice?

No. If your voice is identifiable, that consent is handled separately from the asset licence. Licensing a recording does not by itself grant a right to clone a voice.

Related resources

Want data that clears this in diligence?

Whether you're building a model or sitting on an archive, the first conversation is short and specific.

Send a brief