Voice Cloning Without Consent: What Happens When Someone Takes Your Voice?

What happens when you discover your voice has been cloned, uploaded to an AI platform, and made available for others to use without your permission?

That is the reality voice actor Faye Dicker faced after learning that recordings of her voice had been used to create synthetic versions on a US platform. Her cloned voice had been downloaded more than 900 times before it was removed. In this episode of High Notes, Faye talks to Melissa Thom about the personal impact of that discovery, what she did next and why she believes the issue goes far beyond professional voice actors.

Melissa is also joined by Shannon Sailing, Industrial Official for Audio and Games at Equity, who has worked on Faye’s case and others like it. Together, their conversation moves from the emotional reality of losing control of your voice to practical questions about evidence, contracts, legal protection, consent, and what performers can do when something goes wrong.

Discovering that your voice is no longer under your control

Faye has worked as a voice actor for around 20 years, with credits spanning commercial radio, transport, retail and corporate work. She describes voiceover as something she gradually built into a career, eventually working with brands including Jet2, Matalan, GWR and Boots.

Then one evening, just as she and her husband were about to settle down to watch television, she received a message saying her voice had been cloned.

At first, the scale of what had happened was difficult to take in. The platform offered two versions of her voice, apparently created from transport announcements she had previously recorded. Users could type words and generate speech that sounded recognisably like Faye.

The platform described those voices as available for use across formats including TikTok and audiobooks. Faye had never agreed to that.

What struck her most wasn't simply that the technology could reproduce her voice, but that someone else could decide what it said.

Her husband tried entering something rude to test it. For Faye, that immediately brought home what loss of control really meant. She describes herself as a mother, a woman with values and integrity, and suddenly had no idea what her voice might be saying elsewhere or how it might be being used. She felt sick.

Before asking for a takedown, preserve the evidence

One of the most practical lessons from Faye’s experience is what she did next. Her first instinct might have been to demand that everything be removed immediately, but she chose to document what existed before it disappeared.

She took screenshots, made screen recordings and gathered as much evidence as possible. Only after building that trail did she contact the platform and ask them to remove the cloned voices.

That distinction matters. If a page, voice model, or listing disappears before you record what was there, it may become much harder later to establish exactly what happened.

Faye also connected with other affected voice actors and contacted Equity. The platform eventually confirmed it had removed the material, but for Faye, the problem was already bigger than the takedown itself. Once a voice model has been accessible to other users, you cannot necessarily know who has already downloaded it, generated material with it or passed it elsewhere.

A takedown can stop the immediate problem, but not undo it

Faye is particularly sceptical of relying on fast takedown systems as the answer.

The platform involved in her case later said it had automated its takedown process to under three minutes. But Faye says even one minute is too long if the material should never have been uploaded in the first place.

For every moment a cloned voice is available, she argues, there is no way to know who may be copying it or what they may be making it say.

Her frustration is also about responsibility. Why, she asks, should the person whose voice has been taken have to monitor platforms and prove something shouldn't be there?

For Faye, removal treats what has already happened. It does not address the system that allowed it to happen.

Why the emotional impact matters

The discussion around AI voice often focuses on contracts, technology and economics.

Faye’s story adds something easier to overlook: the personal experience of hearing something that sounds like you, but that you did not say.

After discovering the clones, she wanted to tell everyone what had happened. She describes feeling violated and talks about the experience in terms of her voice, identity and name having been taken.

That sense of identity runs through much of her conversation with Melissa.

A voice is not simply an interchangeable recording. It can carry reputation, personality, relationships and values. If someone can reproduce it convincingly enough, they can make you seem to have said something you never chose to say.

That concern is one reason Faye believes the conversation should not be confined to performers. Anyone whose voice is recorded, whether in a professional studio, a voice note, a meeting or another digital environment, has an interest in understanding what can happen to that material.

What patterns is Equity seeing?

In the second half of the episode, Melissa speaks with Shannon Sailing from Equity, the UK trade union representing performers and creative practitioners.

Shannon works particularly with audio artists, including voice actors and narrators, as well as performers working in video games. Her role includes helping individual members with workplace issues, negotiating terms, and campaigning for wider change.

Faye’s experience is not isolated.

Shannon says Equity is seeing striking similarities between claims. Performers discover their voices on AI platforms despite never agreeing to clone or license their voice for that purpose.

Sometimes the original recording was entirely legitimate. A performer may have been hired to record an audiobook, advertisement or another piece of work. The problem comes later, when that recording is passed on or used for a completely different purpose.

In those cases, the issue isn't necessarily that someone knowingly signed an unfavourable AI agreement. The original consent has effectively been stretched into uses the performer never anticipated and may never have agreed to.

That creates a transparency problem too. Performers may not know where their recording came from, who supplied it, whether it has been used to train an AI model, who has access to that model or where the resulting voice is being used.

What can a union actually do?

When an Equity member brings forward a case, Shannon says the first step is to understand the chain of events.

What did the performer originally agree to? What does their contract say? If there is no formal contract, what was agreed through emails or other correspondence? Who received the recording, and where might it have gone after that?

Equity can then assess possible contractual rights, performers’ rights, copyright, and data protection, and contact the companies involved.

One advantage of the union is that it can see patterns across multiple cases. What looks like a single dispute when one performer approaches a company may look very different when the same behaviour is happening to many people.

But Shannon is also clear about the limitations. In her view, existing UK law does not currently provide a simple route that covers every case. That is one reason Equity is campaigning for legislative change while also trying to improve contractual protections.

Where does your voice sit in UK law?

Shannon describes the legal position of a person’s voice as awkward.

Copyright protects specific works and recordings, but she says the UK currently lacks a comprehensive right that gives an individual control over their voice, image, or likeness in every situation.

Equity has therefore also relied on arguments around biometric and personal data where appropriate. Data protection can help, but Shannon’s wider point is that different laws must address different parts of the problem.

The government has recognised questions around digital replicas and personality rights, but Shannon argues that recognition still needs to become effective legislation.

Consent, control, compensation and transparency

When Melissa asks what effective protection should contain, Shannon reduces it to four principles:

Consent. Control. Compensation. Transparency.

Consent needs to be meaningful. People should understand what they are agreeing to rather than unknowingly signing away broad future uses.

Control means having a say over what can be done with your voice and for how long.

If someone is making commercial use of that voice, appropriate compensation should be part of the conversation.

Transparency matters because performers need to know how their recordings, data, and voice models are used.

Shannon also stresses that these protections apply to more than professional performers. Deepfakes, scams and unauthorised use of somebody’s likeness can affect anyone.

What does success look like after a voice has been cloned?

There is no single answer.

For one performer, success may simply mean getting a cloned voice removed as quickly as possible. For someone else, it may mean compensation or a new agreement covering a use that wasn't originally authorised.

Equity has achieved both kinds of outcome in individual cases.

But Shannon distinguishes between resolving one claim and solving the underlying problem. If a platform removes a handful of voices and then keeps operating the same way, it has treated the symptom, not the cause.

For her, real industry-wide success means performers have clear rights they can rely on, regardless of whether they're working under a particular union agreement.

What should you do if it happens to you?

If performers discover their voice has been cloned or used in a way they didn't agree to, Shannon’s practical advice echoes Faye’s experience.

Don't assume that because something happened, you must have agreed to it.

Preserve the evidence. Find screenshots, recordings and information showing where the voice has appeared. Review your original contracts or correspondence. Try to understand how the recording may have passed from one organisation to another.

For Equity members, Shannon recommends contacting the union so they can help examine the case and the rights involved. She also points to data subject access requests as a way to uncover information an organisation holds about you and potentially understand how data has been shared.

Her message is not to simply accept that once a clone exists, nothing can be done.

The process may feel like trying to put the genie back in the bottle, but challenging unauthorised use also shows that performers and the wider industry don't accept it as normal.

Why this conversation goes beyond voice actors

Both Faye and Shannon repeatedly return to the same point: this issue is bigger than the voiceover industry.

Faye asks why an individual cannot simply own and control their own voice. For her, the need for greater awareness extends from professional performers to family members and the general public.

That wider concern is already feeding into organised campaigns. Equity has been campaigning on AI and performer protection since 2022 through its Stop AI Stealing the Show campaign. Shannon says one key achievement has been shifting the discussion away from treating AI purely as a copyright and training-data issue toward questions of voice, face, likeness, and identity. She also points to the shift away from proposals based on an opt-out system, which Equity had argued would make it easier for technology companies to use people’s data without meaningful permission.  

At the time of recording, however, the campaign was still waiting for the legislative change Equity ultimately wants to see. A promised government consultation on digital replicas and personality rights had not yet opened, and Shannon points out that recognising the problem is not the same as solving it. For Equity, success would mean robust rights that performers and others can actually rely on.

Melissa and Shannon also discuss the newer Save Our Voices Now campaign, which calls for a statutory right for every person to legally own their voice as personal intellectual property. The campaign also sent an open letter to the Prime Minister, signed by well-known actors, including Hugh Bonneville and Matt Lucas.  

Rather than seeing that as competition with Equity’s campaign, Shannon welcomes it. She believes the issue doesn't belong to one organisation, or even to performers alone. When trade unions, actors, public figures, campaigners, and members of the public all make the same point, the pressure grows stronger.

Melissa then connects that directly to one of the wider themes running through this season of High Notes: most people now leave voice recordings in places that have nothing to do with a recording studio. Voice notes, Teams calls, Zoom meetings and recorded workplace conversations all create digital records of a person’s voice.

So while voice actors may be encountering the issue earlier and more visibly because their voices are routinely recorded and commercially licensed, the underlying questions apply much more widely. Who has the right to use a recording of your voice? What did you actually consent to when it was captured? And what happens if that recording is later used to create something you never agreed to?

Can AI and human performance coexist?

Despite everything discussed in the episode, Shannon does not suggest that the future requires eliminating AI voice.

She expects synthetic voices to get better, cheaper, and more widespread, and she accepts that parts of the voice industry will be affected.

But she also sees something fundamentally different in a real performance: a performer's choices, their relationship with the material, and their ability to communicate something genuinely to another person.

She believes AI and human performance can coexist, but only if the performer retains agency.

Her principle is straightforward: if somebody wants to use your voice, they should ask. They should tell you how they'll use it. If they are making money from it, you should be paid. And if you say no, that decision should mean no.

The question is ultimately one of agency

Faye’s experience shows what happens when that agency disappears.

A recording originally created for one legitimate purpose can become the raw material for something entirely different. A recognisable version of your voice can say words you never spoke. And even after a clone is removed, there may be no simple way of knowing where it has already travelled.

Shannon’s work at Equity approaches the same problem from the other side: building cases, strengthening contracts, supporting performers and campaigning for clearer rights.

The four principles she identifies provide a useful framework well beyond the voiceover industry: consent, control, compensation and transparency.

As AI voice becomes easier to create and harder to distinguish from the real thing, the question is not simply what the technology is capable of doing.

It is whether the person behind the voice still has a meaningful choice in what happens to it.

Listen to the full episode of High Notes: Cloned Without Consent, with voice actor Faye Dicker and Shannon Sailing from Equity.