AI video and real people: likeness, consent and deepfake rules

Do not put a real, identifiable person in an AI video without their clear, written permission. That covers their face, body, voice and close lookalikes, and it applies to public figures as much as private ones.

This guide is general information, not legal advice. Rules on likeness, privacy and deepfakes differ by country and sometimes by state or region. For anything beyond your own original characters, check the official source or a lawyer.

A person's face and voice are tied to their identity and reputation. Many places give people some control over commercial use of their likeness, and some have specific rules on deepfakes, especially sexual or deceptive ones. Platforms add their own layer: realistic synthetic content usually needs a label, and some uses are removed outright.

ClickPop does not allow real people's likeness without permission. That rule exists because the harm from a convincing fake can land on someone who never agreed to be in the video.

There is a business reason too. A video that borrows a famous face can be taken down, can lead to a complaint against your account, and can make a brand look careless. Every view it earned disappears with it. Original work keeps its value.

Sometimes the real person is you, a founder or a hired spokesperson. Consent makes this workable, but it should be specific. A good release covers:

Keep the signed release with the project files. If you work with a client, ask for the release before any production starts.

Some uses are off-limits regardless of consent forms or platform. Do not make sexual or intimate content of a real person, content that shows a real person committing a crime or saying something they did not say in order to deceive, fake endorsements, or anything involving minors. Do not imitate a public figure to sell a product.

Lookalikes count too. Prompting for someone who is "basically" a well-known actor, or tweaking a face until it is recognizable, is the same problem with extra steps.

People often think about faces and forget voices. A cloned or imitated voice can identify a person as clearly as a photo, and audio deepfakes are used in scams for exactly that reason. Treat a recognizable voice the same way you treat a face: no use without written consent, and a label wherever platforms ask.

If you need a narrator, pick one of the stock AI voices in the tool, which are not tied to a real person you would be imitating. ClickPop has AI voices in 30+ languages, so you can usually find a voice that suits the brand without copying anyone. If you want your own voice, record it yourself or get it cloned only with your own consent and on a service that allows it.

Never prompt for "a voice like" a named person. Describe qualities instead: "warm, low, unhurried, slight smile in the delivery." You get a voice that fits the brief without borrowing an identity.

For most marketing, storytelling and creator work, an original fictional adult character does the job better. It belongs to your brand, cannot be withdrawn and never ages out of a contract.

Here is a workflow you can follow in ClickPop's Characters studio at /app/characters:

Give the character a name, an age range, a few personality notes and a way of speaking. A host called Mara who is dry, precise and always in olive linen is easier to keep consistent than "a friendly woman". Write those notes into your brief template so every writer and every video uses the same person.

Even a fully consented AI version of a real person should be labeled where platforms ask. As of October 2026, YouTube's altered or synthetic content setting, TikTok's AI-generated content label and Meta's AI info label all apply to realistic synthetic people. In the EU, the AI Act's transparency rules apply from August 2026 and require labeling deepfakes.

For original characters that look realistic, labeling is still the honest choice, and for sponsored posts, the FTC in the US expects virtual influencers to disclose paid relationships the same way people do.