YouTube launches AI deepfake detection tool for all adults

YouTube launches AI deepfake detection tool for all adults. The platform introduces a face scan system and automatic notifications to spot videos that are manipulated by using artificial intelligence.

YouTube has announced a major extension of its AI-generated deepfakes detection system, making it available to all adult users. Up until now, the tool was reserved for certain high-profile creators, journalists, politicians and public figures: it is now available to anyone wishing to protect their digital identity. The aim is to combat the distribution of manipulated videos that exploit a person's face without their authorization.

How the face recognition system works

The system is based on a verification procedure that includes a video selfie and biometric facial recognition. Once the function has been activated, available in YouTube Studio under the section dedicated to "content detection", the platform launches a continuous scan of videos uploaded to YouTube. When the algorithm detects a match between the user's face and suspicious content, it sends an automatic notification allowing the user to intervene.

Notifications and deletion requests: the user takes back control

When a potential unauthorized use of the user's image is detected, the user can analyze the reported content and decide whether or not to initiate a deletion request. Each case is examined on an individual basis, taking into account factors such as the realism of the video, the presence of references to the use of AI and the context in which the face is used. Satirical or parodic content remains excluded from any automatic blocking, and continues to feature among the exceptions provided for in the platform's rules.

Protection limited to the face, not the voice

For the time being, the system focuses exclusively on facial recognition, without extending its attention to voice or other biometric characteristics. This is one of the main limitations of this technology, especially as many deepfakes combine image and audio in order to seem more credible. Access to the service doesn't depend on the number of subscribers or the popularity of the channel, but only on the age of majority and voluntary activation.

Privacy, risks and new challenges for platforms

YouTube has clarified that users will be able to deactivate the service at any time as well as request the deletion of their biometric data. Nevertheless, the large-scale deployment of the system raises new questions linked to privacy protection, recognition reliability and error management, particularly for lesser-known faces. The challenge will be to strike a balance between identity protection and the risk of false alerts or excessive automation.

Source : YouTube Community