Short, for a caption
The one that fits anywhere, including a platform that counts every character.
This video uses an AI version of my face and voice. The words are mine.
DOUBLE CREDITS on your first month, or your first 3 months on annual
Claim nowUsually yes, and the answer gets stricter the more regulated your work is. This is the practical version: what to record before the first video, what to keep on file, what to write where you post it, and the sentences that get people into trouble.
Usually yes. Here are the four things you should be able to point at.
This is education, not legal advice. The rules differ by country, by state and by profession, and yours are the ones that count.
Part of Resources.
Checked September 2026. Free to read, no sign-up, and no gate part way down.

Usually yes, if it is your own face and voice, you have a dated consent record, you label it where you post, and you follow the advertising rules of your profession. The rules bite hardest on regulated work: medicine, law, money. This page is not legal advice. Four things to point at, below.
Who it is for
Anyone whose face is part of their work, and the person who signs off on what the practice posts. Written with regulated work in mind: medicine, dentistry, law, finance.
What you need
A dated consent record for every face in the video, a line to put where you post it, and the advertising or conduct rule that already governs your field.
What it costs
Reading and copying everything on this page costs nothing. A finished talking-head video is 152 credits a second on a paid plan. The Free plan starts you with 8,400 credits and renders up to 60 seconds at a time, and a paid plan runs to thirty minutes. See the plans.
What it does not do
It does not give legal advice, it does not check your video against your regulator's rules, and it does not make a claim safe that would be unsafe said straight to camera.
Search this and you get two kinds of page. A law firm essay about likeness statutes in four US states, and a vendor page telling you AI video is perfectly fine. Neither helps at nine at night with a video rendered and a caption box open.
The reason is that there are two questions here and they get answered as one. There is what you can prove about the video, and there is what you say about it when you post. The first is a filing job you do once, per person, in about ten minutes. The second is one line you paste every time. The rest of this page is those two jobs, in that order.
One thing to get straight before either of them. The AI part does not change what a claim is. A sentence that would breach your field's advertising rules said straight to camera breaches them said by your Digital Twin. The synthetic face is a production method. It is not a shield.
Before you post, you should be able to point at four things. Point at all four and almost every question anyone can ask about that video already has an answer waiting. If you cannot point at one of them, that is the one to fix first, and it is usually the second.
The consent record.
Who agreed, on what date, to what, and how they end it. For your own face that sounds silly right up until a platform, a regulator or a new partner asks who authorised the likeness, and well, it is me turns out not to be a record.
The identity check.
Something that ties the person who gave consent to the person in the picture. This is the record almost nobody has, and it is the one that stops a photo lifted off a website from becoming somebody's Digital Twin. If your tool does not do this, you are the only check. It is the practical half of Likeness rights.
The disclosure.
Two halves. What the file itself carries, and what you write where you post it. The first half is your tool's job. The second half is yours, and it is the half that platforms and regulators actually read.
The field rule.
The advertising standard, bar rule, board guidance or conduct code that already governs what you may say. It existed before AI video and it did not change for it. Look it up once, write the reference into your likeness file, and stop guessing every time.
Here is the honest split, so you know which records you are holding and which ones you still have to make yourself.
The consent recording.
A real person's Digital Twin needs a current consent record before it can render anything. The person says, in their own voice, that this is really them and that they agree to their likeness, voice and mannerisms being used in the videos they choose to make. The Studio issues two digits at the start and listens for them being read aloud, so a clip filmed months ago for something else cannot stand in for consent given today. A consent recording runs between fifteen and ninety seconds.
The identity check, made for you.
Before the consent is accepted, the recording is checked against the photo the Twin was built from. A recording without one clear, matching face all the way through is refused and you record again. That is the second of the four records, and it is the reason a stolen photo does not get very far here.
The expiry.
A consent record runs for one year from the day it was recorded. After that the Digital Twin stops rendering until the person records again. That is on purpose. A permission with no end date is not much of a permission.
Withdrawal.
If the person gave consent through a private link you emailed them, that same email carries a withdrawal link. Using it stops new Renders of them, and it needs no Kyndrify Workspace of their own. The recording itself is kept, because it is the proof that consent was validly given at the time. Withdrawal stops future use. It is not erasure, and you should say so when you ask somebody for their consent in the first place.
The file itself.
Videos and images made on the Free plan carry a visible mark, and the Studio checks for it before it hands the file over. Paid plans download with no visible mark on the file. Built-in AI labelling is still rolling out across output types, so a publisher writes the disclosure line rather than assuming the file carries it. So write your caption as though the file says nothing about itself, because today it might not.
What we do not do.
We hold no certification for any of this and we do not claim one. Nothing in the Studio reads your script against your regulator's rules, so the read-through before you publish is a human one. Kyndrify is not HIPAA-eligible, so patient information does not belong in a Workspace. And Kyndrify does not use your photos, voice, scripts or finished media to train a Kyndrify model.
A free Workspace uses a Starter Twin. A Digital Twin of a named person starts on the Plus plan, and 1080p on the Basic plan and above. Your own voice can be cloned on any plan, Free included. How consent, disclosure and content credentials work at Kyndrify · The biometric information privacy notice
One file per person who appears in a video. Not one per campaign, not one per channel. The consent belongs to the person, so the file follows the person. Keep it wherever your practice already keeps its marketing paperwork, because the point of it is that somebody who is not you can find it.
Copy this, fill it in, date it.
Put the reminder about two weeks before the expiry, not on it.
A Digital Twin that stops rendering on the morning of a launch is an expensive way to learn the date.
Write down who may approve a video in that person's name.
This is the row people skip and the row that matters when a comms lead drafts and a partner signs off. One name, written before the first video, saves an argument after the fifth.
If someone else's face is in the video, their file is not optional.
Get their consent recording before you build the Twin, not after the video exists. You can capture it with them in the room, or send them a private link and let them record it on their own phone in about a minute. Either way the consent record is between you and them. We are the infrastructure, not a party to it.
| Row | What goes in it |
|---|---|
| Person | What goes in itTheir name, and their role in the videos. |
| What they appear in | What goes in itPatient education, firm explainers, weekly posts. |
| Consent recorded | What goes in it20 September 2026 |
| Consent runs to | What goes in it20 September 2027 |
| Re-record reminder | What goes in it6 September 2027 |
| How they withdraw | What goes in itThe link in their consent email, or ask the Workspace owner. |
| Where these videos go | What goes in itYour site, and each social account by name. |
| Disclosure line used | What goes in itPaste the one you chose, below. |
| Field rule we work to | What goes in itYour board, bar or standards reference, and the date you read it. |
| Approved to post in this person's name | What goes in itOne person, by name. |
| Last reviewed | What goes in it20 September 2026 |
| Next review | What goes in it20 December 2026 |
The recording, the identity check and the date all happen in the same sitting, before the first Render.
Whatever the file carries, the caption half is yours. Three lengths, all true, all short enough to survive a character limit. Pick one, use the same one every time, and paste it.
Put the line where it survives a re-share.
A caption stays behind when someone downloads your video and posts it somewhere else. A line burned into the last two seconds of the video travels with it. If the video matters, do both.
Use the platform's own AI-content setting as well, if it has one.
The setting and your sentence do different jobs. Yours explains. The setting is the one the platform enforces, and ticking it is free.
The one that fits anywhere, including a platform that counts every character.
This video uses an AI version of my face and voice. The words are mine.
Where you have room for a sentence about who wrote the words.
This video was made with an AI version of my face and voice. I wrote and approved the script. It is a synthetic recording, not footage of me speaking on the day.
For a regulator, a platform review or a client file, where the names and the dates matter.
This is a synthetic video. It uses an AI Digital Twin of <name>, made with their recorded consent on <date>, and an AI version of their voice. <Name> wrote and approved the script. No part of this is footage of an actual event.
Swap the two bracketed parts for the name and the consent date from that person's likeness file.
Each of these has a fix beside it. None of them is a hypothetical.
Medicine and dentistry.
Patient education is the safe centre of this format: what a procedure involves, what to bring, what happens afterwards. It stops being education the moment it becomes advice for one particular person, so a clinical script closes by naming where to go and handing the viewer back to the clinician who knows their case. Patient information does not go into the Studio. Read next: content marketing for doctors.
Law.
Bar advertising rules differ by jurisdiction, and several ask for a disclaimer on anything that reads like a testimonial or a result. Your firm owns that, not your video tool. And a named attorney's likeness needs two approvals, not one: their own consent record, and the firm's sign-off on what is said in their name. The second is the one people forget. Read next: AI video for lawyers.
Money.
Suitability, disclosure and record-keeping obligations stay with the firm, and no part of this page or of a Render is investment advice. If your compliance team reviews copy before it goes out, then the script is copy and the finished video is the copy going out. Send them the script, not the render, and send it early.
Executives and founders.
Not regulated the same way, and the file matters more, not less. The face belongs to one person and the posting is usually done by a team, so write down who may approve a video in that person's name before the first one ships. Read next: AI video for executives.
If somebody has used your face or voice without your say-so, that is a different door: report it or ask for the content to be removed.
What this page is, and what it is not
This page is education, not legal advice, and reading it creates no professional relationship of any kind. Kyndrify is not a law firm and does not practise law. We hold no certification for consent, provenance or disclosure, and we do not claim one. Kyndrify is not HIPAA-eligible, so protected health information does not belong in the Studio.
The advertising, conduct and record-keeping rules of your field are yours to follow. They differ by country and by state, they change, and no tool makes a video compliant with them. Check yours, write down the date you checked it, and ask your own counsel or compliance lead about anything that matters. If you want to see the obligation in our own words, it is section 5 of the terms.
Checked September 2026. This is a summary, not legal advice.
Build a Digital Twin and the consent record is made in the same sitting: the recording, the identity check and the date, all before the first Render. The Free plan starts you with 8,400 credits.