Posthumous Voice Clone Consent: How to Record Your Wishes
A voice can feel more personal than almost any other digital trace. Photos show what someone looked like. Messages show what they wrote. A recording carries breath, timing, humor, accent, hesitation, and warmth. That is why voice cloning after death can feel comforting to one family member and deeply invasive to another.
Posthumous voice clone consent is the written instruction for whether your recorded voice may be used to generate new audio after you die. It answers a narrow but important question: may someone use software to make it sound as if you said words you never actually recorded?
This is different from preserving real recordings. Keeping a voicemail, podcast episode, interview, video, or family message is archiving. Training or using an AI voice model to produce new speech is a new act. It can create a memorial narration, an audiobook, a documentary voiceover, a song demo, a birthday message, a business announcement, or a fake endorsement. Some uses may be loving. Others may be exploitative, confusing, or legally risky.
The U.S. Copyright Office describes digital replicas as technology that realistically replicates a person's voice or appearance. Its report also warns that unauthorized AI-created replicas can harm private individuals, not only celebrities, performers, and politicians. That matters for estate planning because the question is no longer "Am I famous enough to worry about this?" The better question is "Would I want my family, estate, employer, fans, or strangers deciding what my voice can say later?"
Start with a clear default
Do not begin with the tool. Begin with the rule.
Your instruction should choose one of three defaults: no voice clones, limited private use, or project-specific permission.
No voice clones is the most protective choice. It means your family may save and share real recordings, but no person, company, estate representative, fan community, or business partner may create new synthetic audio that imitates you. This is a good default if you value privacy, dislike simulated interaction, have complicated family dynamics, or simply do not want your voice separated from words you actually spoke.
Limited private use allows a narrow memorial purpose. For example, you might allow a clearly labeled family-only narration made from letters you wrote, or a private archive that helps relatives search existing recordings. This option should never be vague. It needs limits on who may use it, which recordings may train it, whether children may interact with it, whether it may be shared, and when it must be deleted.
Project-specific permission is for a defined creative, educational, commercial, or public purpose. A podcaster may want an estate to finish a nearly complete episode. A performer may want a documentary to use a synthetic line when the estate approves. A founder may want a training course to remain available for a limited time. The key is specificity: one project, one approval process, one contract term, one disclosure rule.
If you are unsure, choose the restrictive default. Survivors can still preserve real audio. What they cannot easily do is pull back a convincing synthetic voice once it has spread through platforms, downloads, reposts, or private copies.
Decide what counts as your voice
A useful consent instruction defines the voice material it covers.
Include obvious recordings: voicemail, family videos, podcasts, interviews, speeches, performances, meetings, voice notes, video calls, livestreams, dictation files, and studio recordings. Then include less obvious sources: social media clips, course videos, customer webinars, smart speaker recordings, game streams, and raw audio files in cloud storage.
For creators and public figures, separate public work from private material. Public podcast episodes or published performances may be appropriate for a licensed project. Private calls, rehearsal files, medical recordings, therapy notes, family videos, and direct messages usually should not become training data.
For families, the emotional line may matter more than the technical one. A three-second laugh in a wedding video, a bedtime story recorded for one child, or a final voicemail may be precious precisely because it is real. Turning it into endless new speech can change the meaning of the memory.
Your instruction can say:
"My real recordings may be preserved and shared by my family. They may not be used to train, create, or publish a synthetic version of my voice."
Or:
"My estate may approve a synthetic voice only for the documentary project already described in my creative files. Private messages, family videos, and medical or legal recordings may not be used."
Plain language is enough. The goal is to make your wishes usable during a stressful moment.
Name who can approve or refuse
Voice clone consent becomes chaotic when several people believe they can decide.
Name one decision-maker. This may be your executor, digital executor, trustee, spouse, adult child, sibling, business manager, creative partner, or another trusted person. The right person is not always the most technical person. It is the person who understands your values, can read a contract, can tolerate pressure, and can say no when a request feels wrong.
Name a backup. If your first choice is unavailable, incapacitated, conflicted, or has died, the backup avoids paralysis.
Then describe conflicts of interest. A person who profits from a voice clone should not be the only person approving it. A business partner may understand the project but have financial pressure. A family member may want a grief tool that other relatives find disturbing. You can require two approvals for public or commercial uses, such as the executor plus a named creative representative.
You can also name people who may not approve a clone. That may feel uncomfortable, but estate planning often protects boundaries by writing them down before emotions run high.
Ban high-risk uses directly
Some uses should be prohibited even if you allow a narrow voice clone.
Most people should ban synthetic voice use for:
- legal instructions or changes to estate wishes
- medical consent or health advice
- financial advice, investment promotion, or debt requests
- political endorsements
- religious or ideological claims you did not personally record
- sexual, romantic, or intimate content
- fundraising that implies a new personal appeal
- customer support or business statements presented as current speech
- messages to children that pretend to be live or responsive
These categories are risky because listeners may treat the audio as your real current intent. A synthetic voice can collapse the difference between memory and instruction. Estate documents, contracts, and family letters should be interpreted from real records, not generated audio.
If you allow any clone, require disclosure. The audience should be told that the voice is AI-generated or synthetically recreated. That disclosure can be spoken in the audio, written in the caption, included in credits, and repeated wherever the file is distributed. The EU Artificial Intelligence Act includes transparency duties for certain AI-generated or manipulated audio, image, or video deepfake content, subject to exceptions. Even outside the EU, the practical principle is sound: do not let people wonder whether the words were really spoken by the person.
Coordinate with creator contracts
Creators need an extra layer of planning.
If your voice has business value, your estate instruction should not sit alone in a folder while your contracts say something else. Review podcast agreements, music agreements, audiobook deals, acting contracts, course licenses, brand deals, talent management agreements, union rules, and platform terms. Some contracts may already mention digital replicas, AI dubbing, synthetic narration, or reuse of recordings.
California AB 1836 addresses specified uses of a deceased personality's digital replica in audiovisual works or sound recordings. Tennessee's ELVIS Act added voice-focused protections to Tennessee personal rights law. Those laws do not replace your planning. They are reminders that voice likeness is becoming a real legal and commercial issue, and the rules can vary by jurisdiction, identity, contract, and type of use.
Creators should leave a simple rights map:
- who manages recordings and masters
- who controls podcast feeds, channels, and archives
- which projects may continue after death
- which projects must stop
- which representatives can negotiate licensing
- whether synthetic voice use is prohibited or allowed only by written contract
- where contracts, releases, and passwords are stored
That map helps families avoid two bad outcomes: losing legitimate work because nobody can act, or approving an overly broad voice license because nobody understands the boundaries.
Give survivors a takedown plan
Your instruction should include what to do if someone creates an unauthorized clone.
Survivors should capture the URL, username, platform, date, screenshots, audio files, captions, comments, and any account information before reporting. They should avoid sharing the clone further while collecting evidence. Then they can use the platform's impersonation, privacy, copyright, harassment, synthetic media, or deepfake reporting channel.
If the clone is commercial, defamatory, intimate, fraudulent, politically manipulative, or widely distributed, the decision-maker should involve an attorney. The estate may need to review publicity rights, privacy rights, contract rights, copyright in the underlying recording, trademark concerns, or consumer protection issues. The exact path depends on the jurisdiction and facts, but a written consent instruction gives the family a clear basis for saying the use is unauthorized.
For public figures and creators, add contact details for the manager, agent, lawyer, label, publisher, platform representative, or business partner who should be notified. For private families, list the person who can speak for the estate so relatives do not send conflicting requests.
Where to store the instruction
Store your voice clone consent with your digital estate plan, not only in a casual note.
Good locations include your estate binder, digital legacy letter, password manager emergency notes, attorney file, creative rights folder, business continuity plan, or trusted family vault. Mention it in your account inventory so the decision-maker can find relevant recordings and contracts.
Review the instruction whenever your voice footprint changes. That might happen when you launch a podcast, record a course, publish videos, sign a talent agreement, become known publicly, start using AI narration tools, or store a large family archive in the cloud.
Conclusion
Posthumous voice clone consent protects more than sound. It protects context, dignity, memory, and trust.
Write the rule before your family needs it: no cloning, private remembrance only, or project-specific permission. Name who decides. Define the recordings that may and may not be used. Ban high-risk contexts. Require disclosure. Leave a takedown process.
Your real voice can remain part of your family's memory and your creative legacy without becoming an open-ended tool for new words you never chose.