HeyGen for Nonprofits: Donor Update Videos and Disclosure Rules

A nonprofit can use HeyGen to make short donor update videos from a script, with an avatar of a staff member who has given explicit consent or a stock presenter. Tell donors plainly that the presenter is AI-generated. Keep sensitive stories, big gifts and thank-yous to major donors on real faces.

The technology is the easy part. A donor who gives $20 a month wants to know their money did something, and a clear 60-second update saying so is worth more than a polished brand film. What goes wrong is trust: a donor who later discovers the "chief executive" in the video was a synthetic copy can feel misled, and for a charity trust is the asset. The sections below cover the workflow, the consent rules, wording you can copy and the cases where a real face is better.

Follow me on Instagram@sagnikteaches

What HeyGen costs a small nonprofit

These are USD list prices from HeyGen's pricing page, checked in October 2026, and they can change.

Connect on LinkedInSagnik Bhattacharya

The Free plan's watermark and one-minute limit make it fine for a test but not for publishing. A nonprofit sending a monthly update will probably sit on Creator. The pricing page doesn't mention a nonprofit discount, and I could not confirm one, so ask HeyGen's sales team before paying list price. For plan comparisons in detail, see HeyGen Creator vs Pro for weekly business videos.

Subscribe on YouTube@sagnikteaches

A quick sum helps the board. A year on Creator at the annual rate (two months free) is $290. If a monthly update video replaces an hour of a staff member's time editing a newsletter or booking a camera, and that hour is worth $25, the plan roughly covers itself in staff time before you count any extra giving. Whether it raises donations is something to test, not assume.

The consent question comes before the script

HeyGen's moderation policy says creating an avatar of another person is prohibited unless you obtain that person's explicit consent. The policy also says the person depicted can ask for removal of their likeness at any time, and that users must honour such requests wherever the content has been published. HeyGen's developer documentation describes a consent step for digital twins, in which the subject records a statement on webcam, though the details vary by account type, so check what your own plan requires.

For a nonprofit, that turns into a short consent checklist:

• Written permission from the staff member or volunteer whose face and voice will be copied, naming the purpose (donor updates) and the period it applies to.
• A clear statement of who can use the avatar, who approves each script, and how the person can withdraw consent.
• A plan for what you do if they leave: delete the avatar, and remove or replace videos where it appears.
• No avatars of trustees, beneficiaries, children or public figures unless each has given that consent on their own behalf.

The consent form is not a legal document by itself. If your charity employs the person, take advice on how this fits with their employment terms, and on whether a likeness licence is needed. For the practical setup steps, the tutorial on HeyGen avatar setup and consent checks walks through creating a digital twin.

Never do this: a realistic mistake

A small charity's communications volunteer builds an avatar of the founder from old fundraising footage "to save her time". The founder isn't asked. The first video goes out, and a long-standing donor writes to say it feels strange that she is thanking people for a gift she never knew about. The charity pulls the video and apologises. The loss is not the video; it is the founder's confidence in her own team. The fix is trivial and must come first: ask, write it down, and let the person approve each script.

Disclosure wording donors will understand

HeyGen's moderation policy doesn't contain a general rule that you must label a video as AI-generated, so the decision falls to you, your regulator and the platform where you post. Rules and expectations on labelling synthetic media differ by place and are changing, and platforms such as YouTube have their own disclosure settings. Take advice if you fundraise across several places. As a matter of donor trust, a plain sentence is cheap and removes any suspicion later.

Wording you can adapt, from lightest to fullest:

On the video, bottom corner:
"Presented by an AI avatar of [first name], with her permission. Script written and approved by the team."

In the email or page text:
"This update was made with an AI presenter built from [first name]'s likeness and voice, with her consent. Every figure in it comes from our [year] accounts. If you would rather hear from a person, reply to this email and [name] will call you."

For a stock presenter:
"The presenter in this video is an AI-generated avatar, not a member of our team. The facts and figures are ours."

Three details make these work. They say who consented, they say who wrote the script, and they offer a human alternative. The last point matters most with older supporters, who may prefer a phone call and are not impressed by being treated as an audience.

Disclosure also covers what is real. If the video shows a field scene, a beneficiary or a building, do not use generated images that look like footage of real people you helped. Mark anything illustrative as illustrative, and use only genuine figures: the total raised, the number of meals served, the date the project opened, all checked against your own records before the script is written.

A donor update workflow that takes about an hour

1. Gather facts (15 minutes). List three real outcomes from the month: a number, a date and a one-line story that the person involved has agreed you may share.
2. Draft the script (10 minutes). Use a general assistant, then edit by hand. Keep it to 150 to 200 spoken words for a 60-to-80-second video.
3. Check every figure (10 minutes). Compare each number with your finance or programme report. An invented or rounded-up number is the quickest way to damage a charity's credibility.
4. Generate the video (10 minutes). Choose the consented avatar or a stock presenter, add a simple background and your logo, and generate.
5. Review frame by frame (10 minutes). Watch with sound for mispronounced names, odd hand movements and any claim you did not write.
6. Add the disclosure and publish (5 minutes). Put the label on-screen and in the text that accompanies the video.

A script prompt that keeps the model honest:

Write a 170-word spoken script for a donor update video from a small charity. Use only these facts and do not add any numbers, quotes, names or outcomes.
Facts: [paste three checked facts]
Tone: warm, plain, no hype, no pressure to give.
Start by thanking supporters without naming individuals. End by telling them where to read the full report. Do not claim donations caused any outcome unless I state it.

A realistic first output says things like "thanks to your incredible generosity, we transformed hundreds of lives". The fixes are to delete the unsupported "transformed hundreds of lives", replace it with the figure from your report ("served 1,240 meals in March"), and cut anything that sounds like a promise of future impact you cannot guarantee.

Translation for supporters who speak another language

HeyGen lists translation into more than 175 languages on paid plans. For a charity with supporters in several language communities, that is a real benefit: the same update can reach people in their first language. It is also a risk, because machine translation can mangle names, place names and the exact meaning of a medical or legal phrase.

Before publishing a translated video, ask a fluent speaker from that community, a volunteer or a bilingual staff member, to watch it with the original script. They should check three things: that the sense is the same, that names and numbers are right, and that the register is respectful and not stiff or slangy. Keep a note of who checked each language and when, so you can correct problems fast. The tutorial on localising a video with HeyGen translation covers the mechanics.

Disclosure still applies in translation. Translate the label too, and say that the translation was machine-assisted and checked by a named role, for example "reviewed by our volunteer coordinator".

When a real face is better

Use an avatar for routine updates and a human for anything where the person on screen is the point. A table of examples helps a board agree rules once.

One illustrative comparison: a small food-bank charity sends two videos in the same month. The monthly update, made with a consented avatar of its coordinator, takes an hour. The thank-you to the families who run a monthly collection is filmed on a phone in the warehouse with a volunteer who speaks for 40 seconds. The second is rougher and does more for the relationship. Do not treat the avatar as a way to avoid showing up.

For turning real event footage into short clips with no avatars at all, see turning one event video into donor clips.

A full example: one month's update for an illustrative food bank

Here is the whole chain for a small food-bank charity, with made-up figures, so you can see where the checks sit. The coordinator has signed a consent form and approved the avatar. The three facts from the March report are: 1,240 meals served, up from 980 in February; a new Thursday collection point opened on 14 March; and the freezer replacement fund reached $3,100 of a $4,500 target.

The model's first draft contains four problems. It says "over 1,200 families were fed" (the report counts meals, not families), it says the fund is "nearly complete" (69% is not nearly), it adds "thanks to our wonderful volunteers" with no figure behind it, and it ends with "every dollar makes a difference", a claim nobody can verify. The edited script keeps the three facts, states the 69% honestly ("we are $1,400 short of the $4,500 target"), and ends with one concrete way to help. The result runs 158 words, about a minute.

The review catches one more error: the avatar pronounces the new collection point's name wrongly. Spelling it phonetically in the script, as a hyphenated word, fixes it. That is the kind of detail only a human watching the video will notice, which is why review stays on the list however good the tool gets.

Making the video easy for every supporter to use

Many donors watch on a phone, with the sound off, or rely on captions. Turn captions on for every video and read them once for errors, especially names and figures. Put the key numbers on screen as text too, so someone skimming still gets the message. If a supporter has a hearing or sight difficulty, offer the script as a short email with the same facts. The text version is also the record of what you told donors, which is useful if anyone later queries a figure.

Keep the video short. Under 90 seconds is a sensible limit for an update, because the donor has already given and wants proof, not persuasion. A longer film belongs on your website for people who want detail.

Stock presenter or custom avatar

HeyGen offers a library of ready-made presenters as well as custom avatars of real people. For a small charity the choice goes like this.

• A stock presenter needs no consent form from anyone on your team and is faster to start with. The cost is that the face is not part of your organisation, so the disclosure wording above ("not a member of our team") is essential, and donors may find it impersonal.
• A custom avatar of a staff member feels more personal and consistent across videos. It needs the consent steps, a clean recording and a plan for the day that person leaves.

Most small charities should start with a stock presenter for two or three updates, collect donor feedback, and only then decide whether a custom avatar is worth the extra consent work. The number of custom avatars each plan includes differs, so check the current pricing page before relying on a figure.

When a donor complains

Expect the odd unhappy reply, and prepare a short, honest response in advance. A reasonable template:

Thank you for telling us. You're right that the presenter in March's video was an AI avatar made with [first name]'s permission, and we should have made that clearer at the start of the email. We have added a note to every update since. If you would prefer to hear from a person, [name] would be glad to call you. The figures in the video came directly from our accounts and you are welcome to see them.

Do not argue about whether AI is a good idea. Agree on what you can: clearer labelling, a human option, and accurate numbers. Log every complaint and review the pattern at your next board meeting. If two or three donors say the same thing, change the practice.

One last habit is worth building in: keep a simple log for each video with the date, the script version, the person who checked the figures, the avatar used and where it was published. When a trustee, auditor or donor asks months later where a number came from, the log answers in a minute, and it shows that a human stood behind every claim the avatar spoke.

Data and privacy rules for scripts and donor lists

Donor information is personal data. Do not paste lists, giving histories, medical details or beneficiary stories into any AI tool unless you have checked its data terms and your own privacy policy, and unless the person has agreed. Use anonymous, aggregate figures: "412 donors gave in March", not names. The tutorial on what donor data must never be pasted into ChatGPT applies to HeyGen scripts just as much, because a script field is a text box like any other.

Personalised videos, such as a thank-you that says "Hello [first name]", need extra care. HeyGen's platform can generate many versions from one template, but check with your data protection adviser first: you are using a donor's name and giving history in a new way. Many small charities are better off with a single general video plus a personal line in the email.

Policies to write before the first video

A single page is enough for a small organisation. It should answer six things:

1. Who may approve a script, and who checks the figures.
2. Which staff or volunteers have consented to avatars, and where the consent forms are kept.
3. The standard disclosure wording and where it appears.
4. Which kinds of message never use an avatar.
5. How long a video stays online, and who removes it if the consent is withdrawn.
6. Who handles a donor complaint about the use of AI.

Boards approve this kind of one-pager quickly, and having one makes the first donor question easy to answer. It also stops a well-meaning volunteer from making an avatar of someone who never agreed.

Judging whether the videos are working

Set measures before the first video. Compare open and click rates on the email with and without the video, replies from donors, and the number of recurring donors who cancel after receiving one. If a video lifts clicks but triggers complaints, adjust the disclosure or return to a real face. After three months, ask five donors directly what they thought. Their answers will tell you more than the dashboard. If you want help setting up a donor update process that suits your team, that is something my AI implementation consultation covers, and you can contact me here.

Questions boards and fundraisers ask about AI avatar videos

Does HeyGen offer a nonprofit discount?

HeyGen's public pricing page doesn't mention nonprofit pricing, and I couldn't confirm a discount. Ask HeyGen's sales team directly before you pay, and compare any offer with the standard Creator plan, which is $29 a month or about $24 a month billed annually.

Do we have to tell donors a video was made with AI?

HeyGen's own policy doesn't require a label, so the decision is yours and your regulators', not the tool's. Rules on AI labelling differ by place and platform. A short plain statement protects donor trust, and the wording below is a reasonable starting point.

Can we make an avatar of our founder who has left or died?

No, not without clear permission. HeyGen's moderation policy requires the explicit consent of the person represented, and a depicted person can ask for removal at any time. For anyone who cannot consent, use a real photo or archive footage instead.

Is it safe to put donor names in the script?

Keep personal data out of scripts unless you have checked HeyGen's data terms and your own donor-privacy rules. Use first names only where donors have agreed, and never paste a donor database into a prompt or script field.

Further reads

• AI Tool Discounts for Charities and Non-Profits: What to Claim — Which AI tools give charities a discount and how to claim it.
• ChatGPT Prompts for Nonprofit Organisations: 10 for Admin and Donors — Ten prompts for admin and donor writing.
• AI Implementation for Nonprofits: A Phased Plan for a Small Team — Where donor videos fit in a phased AI plan for a small team.
• Fireflies for Nonprofits: Can It Write Board Minutes Safely? — Minute the board meeting that approves your AI video policy.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: HeyGen pricing page and HeyGen moderation policy (checked October 2026); HeyGen developer documentation on avatar consent. Prices are USD list prices and can change.