Remember when that fake Drake song went viral last year? Or when unauthorized AI voices of celebrities started flooding the internet? Yeah, that ethical nightmare just got a lot less nightmarish.
ElevenLabs just launched the Iconic Marketplace—a two-sided platform where companies can officially license AI voices of history's most legendary figures. First major partner? Sir Michael Caine himself.
Here's how it works: Instead of random AI companies cloning celebrity voices without permission (the current Wild West situation), estates and rights holders can now offer licensed AI voices through ElevenLabs. Creators submit projects, rights holders approve usage, and everyone gets paid fairly. As Caine put it: "ElevenLabs is using innovation not to replace humanity, but to celebrate it."
The timing's perfect: Academy Award winner Matthew McConaughey just became both an investor and customer, using ElevenLabs to create a Spanish audio version of his newsletter "Lyrics of Livin'" in his own voice. Real celebrity partnerships are replacing sketchy deepfakes.
The marketplace already includes 25+ iconic voices:
- Entertainment legends: Dr. Maya Angelou, Judy Garland, Liza Minnelli, Sir Laurence Olivier.
- Historical figures: Alan Turing, Mark Twain, Thomas Edison, Amelia Earhart.
- Sports icons: Babe Ruth, Rocky Marciano, Johnny Weissmuller.
- Cultural pioneers: Bettie Page, Richard Feynman, Art Garfunkel.
Caine's voice is now available on the ElevenReader app for audiobook narration. But the bigger play here is licensing for films, commercials, games, and other creative projects. ElevenLabs partnered with CMG Worldwide—the agency managing rights for hundreds of deceased celebrities—to make this scalable.
Bonus: ElevenLabs also just added video generation to their studio (powered by Google's Veo 3.1), so you can now create full AI videos with licensed iconic voices. Justine Moore demonstrated this by making a mockumentary of her cat—complete with narration and music—all within one tool. NGL... missed opportunity to make Michael Caine narrate your cat's life, Justine.
Oh, and ElevenLabs also casually dropped Scribe v2 Realtime—a transcription model that might've just killed Whisper, Otter, and every other speech-to-text tool on the market. Sub-50ms latency means it transcribes as you speak (faster than human reaction time), automatically separates multiple speakers without the useless "Speaker 1" labels, and the API is literally three lines of code. Early benchmarks show it outperforming Gemini 2.5 Flash, GPT-4o, and Deepgram Nova 3 across 90+ languages.
Why it matters: This is the first major attempt to make celebrity AI voices work ethically at scale. Instead of deepfakes and lawsuits, we're heading toward a world where estates control how AI versions of beloved figures get used—and get compensated fairly.
Expect other AI voice companies to launch similar marketplaces over the next 6-12 months. OpenAI, Google, and Anthropic (and a whole host of upstarts) are all watching this closely. Meanwhile, ElevenLabs is already signing enterprise customers like Square (AI phone ordering for restaurants) and MasterClass (AI voice clones of Gordon Ramsay and Mark Cuban for personalized coaching).
If you're a creator working on projects that could benefit from a legendary voice, you can now request licensing through ElevenLabs' marketplace and negotiate directly with rights holders. IMO, a version of this model needs to be applied at scale for all IP. If you are a rights holder or someone who wants to license an existing IP, you need an easy platform to connect directly and make the negotiation as seamless as possible. No middlemen needed.
