The AI Voice Cloner OpenAI Won't Release: Here's Why
OpenAI previewed Voice Engine on March 29, 2024—AI that clones voices from 15 seconds of audio. Then decided it was too dangerous to release publicly.
On March 29, 2024, OpenAI showed off Voice Engine—AI that could clone anyone's voice from just 15 seconds of audio.
Then they said: we're not releasing this publicly yet. Maybe ever.
It was one of the rare times an AI company prioritized safety over competitive advantage.
What Voice Engine Could Do
Voice cloning: 15 seconds of audio → full voice replication Any language: Clone voice speaking languages they never spoke Emotional range: Capture tone, emotion, speaking style Quality: Indistinguishable from real person
The demos were impressive and alarming.
Why They Didn't Release It
Election concerns: 2024 was an election year globally Fraud potential: Phone scams, impersonation Consent issues: People's voices used without permission Deepfake proliferation: Adding voice to fake videos
OpenAI concluded the risks outweighed the benefits of public release.
The Selective Testing
OpenAI gave access to:
- Healthcare partners (voice restoration)
- Education companies (language learning)
- Content creators (under strict guidelines)
But no broad public release.
Where Are They Now?
Voice Engine remains in limited preview. OpenAI's caution stands in contrast to the many voice cloning tools that launched publicly anyway.
March 29, 2024 was a rare example of an AI company saying "we built this, but we won't release it"—prioritizing safety over market opportunity.