When AI steals your voice, UK law is silent

A recent news report from June 2026 has highlighted growing concerns that UK law is failing to protect individuals from their voices or faces being cloned by artificial intelligence without their knowledge or consent. Current UK legislation is now considered “unfit for purpose” as it was written before the technology existed. This emphasises the urgent need to bridge the gap between the rapid evolution of AI capabilities and the legal frameworks intended to govern them.

Key issues

Several interconnected concerns have been raised:

• Patchwork legal protection. The UK currently relies on a combination of copyright, data protection, trademarks and passing off, but none of these were designed with AI cloning in mind. Under UK law, there is no general right to own one's voice, image or persona.
• International disparity. France, Italy, the Netherlands and Denmark already have “effective laws” that protect personality rights, and US artists are reportedly being offered larger sums than their UK counterparts for the use of their likeness.
• Government response. While the UK government has acknowledged the issue, stating that it will “launch a consultation to seek views on how we address these harms, while protecting legitimate innovation”, legislation has yet to materialise.

The implications

In June 2026, the University of Cincinnati published research which found that hearing a voice with a similar vocal “timbre” (the unique texture that distinguishes one voice from another) causes listeners’ psychological defences to drop, increasing compliance – even when there is no logical reason to trust the speaker. Criminals need less than ten seconds of audio to clone a voice, and the US Federal Trade Commission has confirmed that imposter scams exploiting this vulnerability have become one of the most common forms of financial fraud.

On the commercial side, AI voice technology is advancing rapidly. Chatterbox, a text-to-speech and voice cloning platform, can replicate a specific voice and then fine-tune tone, pitch, pacing, and emotional register to fit almost any context. Meanwhile, Omilia  launched a native generative TTS model for enterprise contact centres this week that can clone a custom brand voice “in minutes”. While these tools are intended for legitimate uses such as customer service, accessibility, and content creation, they demonstrate how low the technical barriers to voice replication have become.

The current mechanisms for enforcing legal protection may not be sufficient. For instance, Taylor Swift recently filed trade mark applications covering the phrases “Hey, it's Taylor” and “Hey, it's Taylor Swift”. However, trade marks and brand protection strategies only work in commercial contexts and do not solve the broader problem of unauthorised AI imitation. Another possible course of action is passing off, but this relies on demonstrating goodwill or reputation in the voice, which is likely to be available only to very famous individuals, and pursuing it is expensive. Under the current legislative framework, owning a recording or a trade mark does not give you ownership of the sound of your own voice more generally.

A broader look

The growing concern surrounding developments in voice cloning highlights the fragmented nature of the regulatory landscape. More broadly, the debate is shifting from narrow questions of copyright ownership and passing off, towards deeper concepts of identity, dignity and personal autonomy. Companies using synthetic voices in customer-facing applications, marketing or internal training should be reviewing their consent frameworks, vendor contracts, and data governance now, before legislation catches up. The EU AI Act already imposes transparency obligations on synthetic audio, and a UK consultation into digital replicas has been announced for summer 2026.