Quick Answer
Cloning your own voice is legal and, on YouTube, does not require disclosure. Cloning anyone else’s now carries real exposure: the EU AI Act’s transparency rules have applied since August 2026 and bind individuals acting professionally, and several US states have specific voice protections.
Cloning your own voice is legal and, on YouTube, does not even require disclosure. Cloning anyone else's now carries real exposure: the EU AI Act's transparency rules have applied since August 2026 and bind individuals acting professionally, and several US states have specific voice protections. No open-source tool will stop you or warn you.
Key takeaways
- The EU AI Act's Article 50 applies from 2 August 2026 and binds individuals acting professionally, not just companies.
- The federal NO FAKES Act has not passed. US protection comes from state law, which varies.
- The UK's 2026 law covers images only, not voice. This is consistently misreported.
- YouTube does not require disclosure for cloning your own voice for voiceovers or dubs.
- Paid services enforce consent technically. Open-source tools enforce nothing at all.
Cloning your own voice
Start with the straightforward case, because it covers most creators.
Cloning your own voice for narration, dubbing or voiceover is legal and uncontroversial. You hold whatever rights exist in your own voice, so there is no third party to consent.
YouTube goes further and explicitly exempts it from disclosure requirements. Its synthetic content policy requires disclosure for realistic AI content that makes a real person appear to say something they did not, alters real events, or depicts scenes that never happened. Cloning your own voice for voiceovers or dubs is named as exempt.
So a creator using an AI clone of their own voice to narrate videos, or to produce versions in other languages, is on firm ground.
Cloning someone else's voice
This is where 2026 changed significantly, and where most articles are either silent or wrong.
The European Union
The EU AI Act's Article 50 transparency obligations have applied since 2 August 2026. They are in force now, not forthcoming.
Two details matter for creators. First, the obligations bind individuals acting professionally, not only large companies. A freelancer producing client work is covered.
Second, disclosure must be "clear and distinguishable" and perceivable by humans. Embedded metadata alone does not satisfy it: a viewer or listener has to be able to tell.
Artistic and satirical works receive lighter-touch treatment, though not exemption.
The United States
The federal NO FAKES Act has not passed. It was reintroduced in May 2026 and reported out of committee in June, adding a DMCA-style counter-notification process. It is a bill, not law.
Protection comes from state law instead:
- Tennessee's ELVIS Act, effective 1 July 2024, was the first US law to add voice to right-of-publicity protections and covers AI imitation directly.
- California AB 2602, effective 1 January 2025, requires specific consent for digital voice replicas in contracts.
- California AB 1836, effective 1 January 2026, covers deceased personalities, requiring estate consent with damages starting at $10,000.
The United Kingdom
The UK position is the most misreported.
The Data (Use and Access) Act 2025, section 138 came into force on 6 February 2026, criminalising both the creation and the requesting of non-consensual intimate images.
It covers images only. Not voice, not audio. Numerous articles describe it as a voice-cloning law. It is not.
The UK has no dedicated voice-cloning statute. Recourse runs through passing-off, data protection or fraud law, and commentators have described the existing framework as unfit for purpose.
Denmark and beyond
Denmark is introducing a neighbouring right over voice and appearance, requiring consent and lasting 50 years after death, expected in force July 2026, enforced through the Digital Services Act. Parody and satire are excluded. Ireland has signalled it may follow.
The European Parliament's own briefing notes these rules "do not constitute copyright in the strict sense," several member states object, and the effect is limited to Danish territory through geo-blocking.
What the platforms require
YouTube requires disclosure for realistic synthetic content in the categories above. Non-disclosure risks removal or suspension from the Partner Programme. Critically, disclosure does not affect monetization or reach. YouTube states it "won't limit a video's audience or impact its eligibility to earn money."
TikTok requires labelling of realistic AI content and goes further: synthetic media of real private individuals is banned outright, even when labelled. It has applied C2PA automatic labelling since January 2025.
Spotify permits AI music but states that "vocal impersonation is only allowed in music on Spotify when the impersonated artist has authorized the usage."
The enforcement gap that matters
Here is the practical difference between paid and free tools, and it has nothing to do with audio quality.
ElevenLabs requires a recorded consent statement spoken in the target voice before professional voice cloning. That is a technical gate. You cannot clone someone without their participation in the process.
Open-source tools enforce nothing. voicebox ships a responsible-use policy requiring you to clone only your own voice or one you have explicit permission to use, prohibiting impersonation, fraud and commercial use of a person's voice without legal right.
That policy is a text file, not a control. The software runs on your machine with no gatekeeper. It will clone whatever audio you feed it and never ask whether you had permission.
The entire legal responsibility transfers to you. That is not an argument against open-source tools. Our guide to the best free AI voice generators covers why voicebox and GPT-SoVITS are excellent. It is an argument for knowing what you are taking on.
Practical guidance
Cloning your own voice: proceed. Legal, and exempt from YouTube disclosure.
Cloning with permission: get consent in writing, specifying the uses permitted. California's AB 2602 makes specific consent a contractual requirement, which is a reasonable standard to adopt anywhere.
Cloning a public figure: do not, unless authorised. Tennessee and California provide direct causes of action, EU transparency rules apply, and platforms have their own bans.
Cloning a deceased person: California AB 1836 requires estate consent with statutory damages. Denmark's rule runs 50 years post-mortem.
Client work in the EU or UK: the EU AI Act binds you as a professional. Disclose clearly and in a way a human can perceive.
This is not legal advice. Law here is moving quickly and varies by jurisdiction. For anything commercially significant, take proper counsel.
The verdict
The useful summary is short: your own voice is fine, anyone else's needs consent, and the law caught up in 2026 in ways most articles have not registered.
The gap worth internalising is that free tools give you capability without guardrails. A paid service asking for a recorded consent statement is not being obstructive: it is doing something the open-source alternative structurally cannot. When you choose the free tool, you also choose to be the compliance layer.
Related reading: voicebox review and how to clone your voice with AI.
Pricing verified against each vendor's own pricing page in September 2026. Plans change often, so check the vendor's page before you buy.
Frequently Asked Questions
Is it legal to clone your own voice?
Yes, and it is uncontroversial. You hold whatever rights exist in your own voice, so there is no third party to consent. YouTube explicitly exempts cloning your own voice for voiceovers or dubs from its synthetic content disclosure requirements.
Is it legal to clone someone else's voice?
Only with consent, and the law tightened in 2026. The EU AI Act's Article 50 has applied since 2 August 2026 and binds individuals acting professionally. Tennessee's ELVIS Act and California's AB 2602 and AB 1836 create US state protections. Cloning a public figure without authorisation is not advisable.
Does the UK have a voice cloning law?
No dedicated statute. The Data (Use and Access) Act 2025 section 138, in force February 2026, covers non-consensual intimate images only, not voice or audio. This is consistently misreported. UK recourse runs through passing-off, data protection or fraud law.
Has the NO FAKES Act passed?
No. It was reintroduced in May 2026 and reported out of committee in June, but it remains a bill rather than law. US protection currently comes from state legislation, which varies by state.
Do free voice tools check for consent?
No. Open-source tools like voicebox ship responsible-use policies requiring permission, but a policy is a text file, not a technical control. ElevenLabs by contrast requires a recorded consent statement spoken in the target voice before professional cloning. With free tools the entire legal responsibility is yours.