YouTube adds voice detection to likeness tool
YouTube is expanding its likeness detection technology to include voice analysis, aiming to better protect creators from AI-generated voice clones. The feature will also become available on mobile devices.
YouTube announced on September 23 that it is incorporating voice detection into its likeness detection tool, in addition to its existing facial recognition capabilities. This expansion aims to provide creators with enhanced protection against the unauthorized cloning of their voices and likeness.
The company stated that the new voice detection feature will improve overall match accuracy and lay the groundwork for broader voice protections in the future. The likeness detection tool will also be accessible through the YouTube mobile app, allowing creators to enroll, receive match alerts, and take action directly from their smartphones.
This development comes amidst increasing legal scrutiny of AI voice cloning in countries like India, where courts have issued injunctions against unauthorized voice cloning for public figures. While YouTube's tool currently focuses on improving face match accuracy, the inclusion of voice detection addresses concerns about audio deepfakes. The platform has not yet specified when voice detection will be available in India.
To use the service, creators will need to enroll by providing a government-issued ID and a selfie video. YouTube will store this data, along with voice templates, for up to three years. The platform reserves the right to review flagged content and will consider factors such as parody and satire before taking action.
Likeness detection was initially launched for YouTube Partner Program members in October 2025 and has since been extended to journalists, public officials, and celebrities. The addition of voice detection marks a significant step in addressing the challenges posed by AI-generated content, although questions remain regarding its full capabilities and regional rollout.