Where to find licensed faces for advertising (and why stock photos don't work)

By HyperKnown TeamPublished September 2, 2026Updated September 24, 2026

Company enquiries about AI advertising tend to arrive as two questions: do you have a catalog of licensed faces we can browse, and what does one cost. Fair questions with surprisingly few good answers, because most of what a search for faces returns is stock photography, which does not carry the rights this use case needs, and most genuine likeness providers keep their rosters private behind a sales call.

This guide covers where licensed faces actually come from, why the stock-photo shortcut creates legal exposure instead of saving money, and what a trustworthy catalog should show you before you spend anything. What drives the cost is covered in its own guide; HyperKnown quotes each brief as a flat, all-in fee per person for a defined usage.

Where can companies find licensed faces for advertising?

The differences show up in what you can inspect before committing. A marketplace like HyperKnown works like a catalog: you browse verified people, see each person’s recorded content restrictions, and send a brief the person accepts or declines. You pay nothing unless they accept. Agencies and studios can reach talent a marketplace cannot, but the roster is private, the likeness cost is folded into a production quote, and the consent paperwork is whatever that shop negotiated. Avatar libraries are the fastest and cheapest option and the least equivalent: the face is either fully synthetic or licensed from a person under a one-time buyout years before your campaign existed.

Can I use stock photos for AI-generated ads?

The distinction is between copyright and likeness rights. The stock agency licenses you the image file; the person's face is covered only as it appears in that exact photograph, under whatever release they signed at the shoot. Generate a video of that person saying words they never said and you are outside the release, into right-of-publicity territory, where most US states let a person sue over commercial use of their identity without consent, and a growing set of digital-replica statutes applies specifically to AI-generated likenesses. The AI likeness law map covers those rules by jurisdiction.

So the search term "stock faces" leads somewhere real but wrong: royalty-free portrait libraries solve for illustration, not for AI advertising. The AI-era equivalent of the model release is a likeness license granted for your specific campaign, by a person who can see what you intend to make.

What should a licensed-face catalog actually show?

A catalog is only as good as what each listing lets you verify. Before licensing a face from anywhere, check for six things:

1. Verified identity. The platform should have verified that the person behind the profile is who the photos show, not just accepted an upload.
2. A clear fee basis. You should know how the fee is calculated before you brief anyone: on HyperKnown, a flat, all-in fee per person for a defined usage, with the person receiving 100% of their agreed fee.
3. Explicit channel rights. The license should name where the content may run: social, paid digital, broadcast TV, CTV/OTT, out-of-home. Silence about channels becomes an argument later.
4. Documented permission. Check individual project approval or a separate programme opt-in covering the proposed use. Neither a public roster nor platform membership alone establishes commercial permission.
5. A consent record you receive. Documented evidence tying the person, your brief, the terms, and their approval together, ready for platform review or procurement.
6. Provenance on the output. Content credentials such as C2PA let the finished ad carry a verifiable record of how it was made and under what license.

How do avatar libraries differ from licensed real people?

Avatar libraries are legitimate for what they are; the risk is treating them as interchangeable with licensed real people. A synthetic face cannot vouch for a product, carries no audience, and in several jurisdictions now triggers AI-disclosure labeling that changes how the ad reads. The performance and legal trade-offs are unpacked in real people vs synthetic avatars in AI ads. When the campaign calls for a real, recognizable, consenting human, the sourcing question reduces to one test: can this provider show you the person's consent to your campaign, on paper, before you pay.