Voice cloning, consent and the law: what creators need to know

October 22, 2026 · 7 min read · By Resona Editorial Team · Resona Blog

Voice cloning is a neutral technology — the ethics live entirely in how it is used. This article covers the one rule that matters most (consent), the situations that get creators into trouble, and the disclosure habits that keep synthetic audio trustworthy. It is general information, not legal advice; rules differ by country and change quickly.

1. Consent is the baseline, not a nice-to-have

Every responsible use of voice cloning starts from the same question: does the person whose voice is being cloned agree to it, knowing what it will be used for?

Your own voice — fine. You are the rights holder.
A voice actor or narrator you hired — fine, if the contract says the synthetic voice may be created and used for the intended purpose. Get this in writing, and be specific about scope (which projects, how long).
A friend or family member — fine for personal projects if they genuinely understand the output will speak words they never said.
Anyone else — celebrities, colleagues, people in recordings you found — not fine. "Technically possible" is not permission.

If you cannot point to the person's agreement, do not clone the voice. There is no workflow that fixes a missing consent retroactively.

2. The three situations that cause real problems

Impersonation and fraud. Cloning someone's voice to deceive — fake emergency calls, fraudulent authorizations, bogus executive instructions — is the use case regulators and courts care about most, and it is treated as fraud or impersonation in essentially every jurisdiction. The synthetic origin does not soften it.

Non-consensual intimate or defamatory content. Synthetic audio that puts a real person's voice into sexual, humiliating or false statements causes direct, provable harm. Platforms (including ours) treat this as prohibited content, and several countries have written synthetic media specifically into harassment and defamation law.

Commercial voice likeness. A recognizable voice can function like a trademark or personality right. Using a distinctive voice to imply endorsement — in an ad, a product demo, or political content — can infringe even when no clip was copied.

3. Disclosure: the habit that protects everyone

Rules on labelling synthetic media are tightening — the EU's AI Act, various national deepfake laws, and platform policies all push in the same direction: tell the audience when a voice is synthetic, especially in news, political, or commercial contexts.

Practical habits that cost nothing:

Say it in the content: a short "AI-generated narration" line in a video description or podcast intro.
Say it visually: a small label on screen during synthetic segments.
Keep the consent records: who agreed, to what, when. If a question ever arises, the paper trail is the answer.

Disclosure also builds trust. Audiences are learning to distrust "too perfect" audio; a creator who labels synthetic narration looks more credible, not less.

4. What platforms are responsible for

Serious services do not just leave this to users. Reasonable baseline you should expect from any voice platform — and what Resona enforces:

Prohibited-use policies that explicitly ban impersonation, fraud, and non-consensual content, with account termination for violations. See our terms of service for the full list.
Content screening on inputs and outputs — machine checks plus human review — covering the chain from text you type to audio you download.
Reporting channels so affected people can flag abuse and get a response.

If a platform has none of these, that is a signal about how it will handle your data too.

5. A simple decision test

Before generating, ask three questions:

Would the person whose voice this is be comfortable if they heard the result, knowing how it will be used?
Could a reasonable listener mistake this for the actual person speaking, in a context where that matters?
If this audio were published with my name on it, would I be comfortable explaining how it was made?

If any answer gives you pause, change the plan — different sample, clearer label, or written permission.

What consent do you need to clone an employee's or an actor's face and voice?

To clone an employee's or an actor's voice with Resona, you need their written, informed consent that names the exact project and how the synthetic voice will be used. For employees, the agreement should come from the person whose voice is recorded — and, where an employment contract assigns voice rights to the employer, from the employer as well. For actors and narrators, consent lives in the booking or contract: get the synthetic-voice usage scope in writing, including which projects and for how long. "Face" cloning follows the same baseline — any likeness used to speak or appear needs the person's permission first. If you cannot point to that agreement, do not generate the clone. Read the terms of service for prohibited-use rules, or open the console to start a clone. Everything runs on Resona — pay per use, no subscription.

The short version

Clone voices you have rights to, label synthetic audio when context makes identity matter, and keep records. Everything else — quality, cost, workflow — is easy compared to getting consent wrong.

With consent handled, the practical side is quick: upload the sample in the Resona console, or start from a licensed library voice. Cloning is 100 credits (about $2); current rates are on the pricing page.

New to this topic? Read the full Voice Cloning Guide.