AI text-to-speech programs could “unlearn” how to imitate certain people
MIT Technology Review - AI
Jul 15, 2025 10:00
Peter Hall
1 views
airesearchtechnology
Summary
Researchers are developing "machine unlearning" techniques to make AI text-to-speech models forget how to imitate specific voices, aiming to combat the misuse of audio deepfakes for fraud and scams. This approach addresses growing concerns about the ethical risks of advanced voice cloning technology and could help protect individuals' vocal identities in the evolving AI landscape.
A technique known as “machine unlearning” could teach AI models to forget specific voices—an important step in stopping the rise of audio deepfakes, where someone’s voice is copied to carry out fraud or scams. Recent advances in artificial intelligence have revolutionized the quality of text-to-speech technology so that people can convincingly re-create a piece of…