YouTube will start combining speaking-voice detection with facial detection in its likeness protection tool later this year, the company said at its Made on YouTube event on September 23. The change is positioned as an accuracy improvement, not a new product line.
Aparna Pappu, Vice President at YouTube, wrote in a blog post: “But likeness is not just the way you look, it’s also the way you sound.” That line sets up the update. What follows, though, is a list of open questions about voice-only clones.
What the tool does today
The existing tool scans newly uploaded videos for the faces of enrolled creators only. If a creator finds an AI-altered version of their voice, YouTube’s help page says they must report it manually through the privacy complaint process.
- Enrolling means consenting to audio processing; YouTube may later ask for a reference voice sample.
- Enrolment requires a government-issued ID and a selfie video.
- YouTube stores the selfie video, legal name and likeness template for up to three years from the last sign-in.
- An optional checkbox lets YouTube use creators’ face and voice templates to develop and improve its likeness detection models.
- Not every flagged video is removed; parody and satire are considered.
Voice-only clones stay a gap
YouTube described voice detection as a way to improve face matches. It has not said whether the tool will flag a cloned voice when the creator’s face does not appear. That is the scenario at the centre of Indian singers’ cases. The help page also calls likeness detection experimental and says it is unavailable in some countries, with no timeline for India.
The company will bring likeness detection to the YouTube mobile app, so creators can enrol, get match alerts and act from their phones. That makes the tool easier to use, but it does not answer the voice-only question.
Why it matters for Indian creators
Indian courts have repeatedly restrained AI voice cloning, often through orders that bind YouTube. In August 2024, the Bombay High Court restrained unauthorised use of Arijit Singh’s name, voice, image and likeness. In October 2025, it restrained cloning of Asha Bhosle’s voice and directed platforms including YouTube to remove infringing videos. Similar orders have followed for actors, politicians and public figures.
Those protections generally go to people who can prove celebrity status and afford litigation. Most Indian creators have neither. An upload-time detection tool would let them catch clones before they spread. As announced, the voice feature only sharpens face matches, so cloned songs and voiceovers remain outside it.
Compliance context
The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026, effective February 20, require platforms to deploy “reasonable and appropriate technical measures” against synthetically generated information that falsely depicts a person’s identity or voice. SFLC.in told MediaNama that the rules may not apply where a person’s likeness or voice is used beyond consent unless the content is also misleading or otherwise unlawful.
YouTube has not said whether it counts likeness detection as such a measure. MediaNama has sent YouTube questions on voice-only clones, India availability, court-ordered removals, generative AI training and the IT rules.
Source: MEDIANAMA
