AI spokesperson videos: how they are made and when to use one

An AI spokesperson is a generated on-screen presenter who delivers your script to camera, in your brand’s style, without a shoot. There are two ways to get one: a digital double of a real person who has agreed to it in writing, or an invented presenter designed from scratch. It works well for explainers, product updates and multilingual versions, and it is the wrong choice for testimonials or anything that depends on trust in a real human being.

What is an AI spokesperson?

It is a consistent face and voice that presents for a company across many videos. At one end are stock avatars inside a subscription tool: pick a face, type a script, download a talking head. At the other end is a custom presenter built for one brand and used like a brand character, with the same face, voice, wardrobe and manner in every video.

Continuity is what separates a spokesperson from a one-off generated person. By the third video, viewers recognise them. A stock avatar is recognisable too, unfortunately as the same face that presented a competitor’s product last week.

The two routes: a digital double or an invented presenter

	Digital double	Invented presenter
Based on	A real person: a founder, an employee, a hired actor	Nobody. Designed from a brief
What you need first	Written consent and a recording session	A character brief and rounds of still images
Strength	Carries a real person’s credibility, and that person can approve every script	Nobody can leave the company, change their look or withdraw permission
Main risk	The person leaves or changes their mind	Looking like a real person by accident, or being mistaken for one
Typical use	Founders, in-house experts, trainers	A lasting brand face, ads at volume

At Zero Shot, an AI video studio in London, we build both kinds of AI presenter. We do not create a likeness of anyone who has not agreed to it.

How an AI spokesperson video is made, step by step

1. Define the job. Decide what the presenter is for, who they speak to, on which platforms and in which languages.
2. Get consent, or design the character. For a digital double, the agreement is signed first, then the person records clean footage to camera in good light, plus clean audio. For an invented presenter, the work starts as still images: face, age, wardrobe and setting, revised until you approve one look. That look becomes a reference sheet that every later shot is checked against.
3. Choose the voice. There are three options: a human voice actor, a clone of the real person’s voice made with their consent, or a licensed synthetic voice. Pick one and keep it.
4. Write for speech. Short sentences, one idea each. Read the script aloud with a timer before anything is generated.
5. Generate the performance. The talking shots are produced in short takes and checked for mouth shapes, teeth, blinking, hands and eye line.
6. Edit like a normal video. Cutaways, product shots, captions, music and sound. A presenter talking for forty seconds without a cut is dull whether filmed or generated. The stages match how AI reels are made.
7. Check claims and labels. Every factual claim in the script is checked, the platform’s AI label is applied where required, and the files are exported for each placement.

Where does an AI spokesperson work well?

• Explainers and how-tos. A steady presenter walking through a feature or a process.
• Product updates. When the feature changes, one line is regenerated. Nobody rebooks a studio.
• Multilingual versions. The same presenter can deliver the same script in English and in Russian, with lip movement matched to each language.
• Onboarding and training. Internal videos date quickly, and updating one means editing the script.
• Ads at volume. Twelve openings for one offer, tested against each other. We cover this format in our guide to AI UGC ads, and it is the core of our AI ads service.

Where does it not work?

• Testimonials. A generated person has never used your product. There is nothing honest for them to testify to.
• Anything that rests on a real person’s trust. A doctor’s advice, a financial adviser’s opinion, a founder’s apology. If the message is “believe me”, it needs a real “me”, or at least a consented double of one whose words that person has approved.
• Real places and real events. Your restaurant, your clinic, last night’s launch. Film them. Our guide to choosing between AI video and a shoot goes through the cases.

Consent and likeness: what needs to be in writing?

For a digital double, the agreement with the person should cover:

• what is captured: face, voice or both
• where the videos may run: organic, paid advertising, which countries, which languages
• who approves each script before release
• how long the permission lasts, and what happens if the person leaves the company
• how they can withdraw, and how quickly videos come down
• where the source recordings and any trained models are stored, and when they are deleted
• subjects that are off limits

The UK has no single “image right”. Protection comes from several places at once. CAP Code rule 6.1 urges marketers to get written permission before portraying a member of the public, referring to a person with a public profile, or implying that someone personally approves of a product. A recording of a person’s face and voice is also their personal data. In the US, the right of publicity is state law and differs from state to state.

For an invented presenter, two things matter: that they do not resemble a real person, famous or otherwise, and that the contract says who owns the character. This is general information and not legal advice, and a contract that involves someone’s likeness is worth a solicitor’s time.

What are the disclosure and advertising rules in the UK and the US?

United Kingdom. The CAP and BCAP Codes contain no AI-specific rules. The existing ones apply however the ad was made. In guidance published in May 2025, the ASA said there is no blanket legal requirement to disclose AI in UK ads. It gave advertisers two questions: is the audience likely to be misled if the use of AI is not disclosed, and does the disclosure clarify the ad’s message or contradict it? It also warned that disclosure alone is very unlikely to fix a fundamentally misleading message. Separately, CAP Code rule 3.47 requires marketers to hold documentary evidence that a testimonial is genuine. A generated person cannot supply any.

United States. The FTC’s rule on consumer reviews and testimonials, in force since October 2024, covers reviews and testimonials that misrepresent that they are by someone who does not exist, “such as AI-generated fake reviews”, or by someone with no actual experience of the product. The FTC can seek civil penalties against knowing violators. The FTC’s Endorsement Guides, revised in 2023, were also updated so that the definition of an endorsement takes in virtual influencers.

European Union. If your videos run in the EU, the AI Act’s transparency obligations have applied since 2 August 2026, and they include telling people when they are shown a deepfake. Check how that applies to your presenter before a European campaign.

The practical line is simple. An AI spokesperson saying “here is how the app works” is a presenter. An AI spokesperson saying “I have used this for six months and my skin cleared up” is a fabricated testimonial, and a small “made with AI” caption does not repair it.

Do platforms label AI spokesperson videos?

Yes, and the rules differ by platform.

• Meta. Its help centre says that when an ad contains an AI-generated photorealistic human, the “AI info” label appears next to “Sponsored” at the top of the ad, not inside the three-dot menu. Meta describes this for ads made or materially edited with its own generative tools, and says the rollout is gradual.
• YouTube. Creators must disclose realistic altered or synthetic content, including anything that makes a real person appear to say or do something they did not. That describes a digital double exactly.
• TikTok. Its policy requires people to label AI-generated content that contains realistic images, audio or video.

These policies change often, so read the current version before each campaign. Our default is to apply the label whenever a realistic generated person is on screen.

Should you commission one?

If you explain things for a living, publish often and work in more than one language, an AI spokesperson is worth serious thought. If your marketing depends on real customers and real results, keep those people on camera and use a presenter for everything around them. Our checklist for choosing an AI video production company lists what to ask a supplier about consent and ownership. If you have not commissioned generated video before, our guide to what to expect from an AI video agency in London follows a project from brief to delivery.

To talk through a presenter for your brand, send us a short brief by WhatsApp or email. Tell us what the presenter needs to explain and in which languages, and we will reply with a fixed price.