YouTube requires facial biometrics against deepfakes, but policy allows AI to be trained with data

YouTube launched a tool in October that allows creators to request the removal of videos with deepfakes that use their face or voice. Para To access the resource, the platform requires an official document with a photo and a facial biometric verification video. The functionality began to be released gradually and should reach more than 3 million participants from Programa to Parcerias by January 2026.
Privacy experts and image protection companies point out that biometric data sent is subject to Política of Privacidade of Google. Esse document authorizes the use of public information, including facial characteristics, to improve products and train artificial intelligence models. Embora YouTube states that it never used this specific data for training, the future possibility remains open.
The controversy gained strength after creators and specialized companies declared that they do not recommend using the service in its current format.
How the detection system works
The “similarity detection” tool automatically identifies content that synthetically reproduces the creator’s face or voice.
The process requires the sending of a short video where the person repeats phrases displayed on the screen, in addition to an official document.
These materials, according to YouTube, only serve to confirm identity and enable the removal feature.
Concerns raised by experts
Companies like Vermillio and Loti, specialized in image rights protection, advise their customers to avoid the tool.
They argue that voluntarily submitting high-quality biometrics can facilitate future training of generative AI models.
Política of Privacidade of Google, which YouTube links to, explicitly allows the use of public facial data for product development.
The document does not distinguish between information collected in different company services.
Official position of the platform
In response to CNBC, YouTube stated that it never used biometric data from this tool to train artificial intelligence systems.
The company said it will review the wording of the program for greater clarity, but will maintain the current privacy policy.
The company is also studying forms of financial compensation for creators affected by unauthorized use of images, with no defined deadline.

Experience of affected creators
Doctor and influencer Mikhail Varshavski, known as Doctor Mike, reports finding deepfakes featuring his image in fake advertisements every week.
He highlights that tools like Sora (OpenAI) and Veo 3 (Google) have made creating synthetic videos more accessible and realistic.
Other creators report a significant increase in manipulated content since the public launch of major video generation models.
Alternatives adopted by the market
Some creators opt for third-party deepfake monitoring and removal services that don’t require submitting new biometric data.
Specialized platforms offer constant internet scanning and automated legal action against violations.
These solutions, however, have a monthly cost and limited reach compared to YouTube’s scale.
Gradual feature expansion
Full access to the tool will be granted in phases until January 2026 to all eligible members of Programa of Parcerias.
The platform reported that more than 3 million channels should receive the functionality in the coming months.
YouTube continues to adjust eligibility criteria and review processes for takedown requests.
The case highlights the growing conflict between combating deepfakes and protecting personal data in the era of generative artificial intelligence.






