Tool called Voice Engine can clone voices with just 15 seconds of audio, too risky for general release due to misinformation threats.
OpenAI is taking a cautious approach to releasing Voice Engine widely, focusing on responsible deployment and societal adaptation.