When Italian Prime Minister Giorgia Meloni moved last week to trademark her voice, she underscored the rising concern over deepfakes: increasingly convincing synthetic voices that can mimic politicians, CEOs and other public figures. Her action also drew attention to the rapidly expanding business of helping prominent people secure rights to their voice likenesses and to companies that argue voice AI will become a primary interface for customer service, sales calls and interactions with devices from cars to robots.
Valuation, clients and product use
London-based ElevenLabs became one of the most highly valued startups in the voice sector this month after a $300 million employee tender that implied a $22 billion valuation. The company has worked with celebrities including Matthew McConaughey and Sir Michael Caine, and created a custom voice for Yahoo based on co-founder Jerry Yang. ElevenLabs says it has developed comprehensive safeguards to detect and prevent unauthorized voice cloning and other abuses.
Most of the company’s revenue comes from corporate customers such as Deutsche Telekom and Revolut. Co-founder Mateusz “Mati” Staniszewski believes businesses should prepare for a future in which customers increasingly expect to speak with AI agents rather than type to chatbots.
A concrete example: voice for recruiting
Staniszewski gave a concrete use case: an AI interviewer for recruiting sales and business development representatives. Instead of filling in a web form, candidates can speak with an agent. The switch increased engagement in two ways: more people applied because speaking is quicker than typing, and candidates provided more information in conversation. To date, the company says it has received more than 3,000 leads through this channel, of which over 500 were qualified.
Staniszewski emphasises that for an AI agent to work it needs the right knowledge base, workflows, and a period of coaching in early phases. He argues companies should not set humans and agents in competition but should have people manage agents and be fairly rewarded for doing so. It should be an iterative process with an agent–human–agent loop, and the AI must know when to bring in a human.
Safety, rights and market-building
ElevenLabs says users can always see how an AI voice was generated and that there are mechanisms to take action if fraud occurs. The company notes that requests involving money trigger blocks or flags for human review, and that producing a high-quality voice clone requires additional authentication steps, so cloning a voice is not easily or automatically done.
The company also offers a marketplace-style approach: people can record and authenticate their voices, share them, and earn compensation. That model helps explain why artists such as Sir Michael Caine have chosen to work with the firm rather than oppose voice-AI developments — they can participate commercially if their rights are respected.
IPO plans and long-term ambitions
Staniszewski said ElevenLabs would like to go public one day and hopes to be ready within 18–24 months, leaving the timing of any listing to wider market triggers. He cited the desire to let users and partners share in the company’s journey as one reason to consider a future public offering.
European origins as an advantage
Staniszewski views the company’s European roots — founded in Poland and now based in London — as a major advantage. London provides ample talent eager to build ambitious global companies, and because ElevenLabs focused from the start on a polyglot European market, creating a model that works across many languages was never an afterthought.
Conclusion
Voice AI is rapidly maturing into a commercial technology with measurable benefits for customer engagement and operational use cases, while legal and ethical issues around unauthorized cloning, creators’ rights and fraud prevention remain central. ElevenLabs’s high valuation, corporate client base and celebrity partnerships position it as a leading player as the sector seeks scalable safeguards and long-term business models.
