YouTube is preparing to take action to provide its creators with tools to protect their content from being used as the basis for deepfakes, highly realistic fake videos produced with the help of artificial intelligence.
While artificial intelligence offers many new possibilities to online creators, it can also be a vector for content being used and reused without the creator's authorisation. This is why Google, via YouTube, is currently developing safeguards to address the concerns of its creators (but also artists), with tools designed to combat the unauthorised use of content made available on the platform.
First of all, YouTube will soon be integrating a new technology for identifying synthetic songs within Content ID, its copyright protection system. This system enables content creators, record companies and movie studios to identify whether their works are being used on YouTube without their authorisation. A tool will soon be added to identify content generated by AI simulating a person's singing voice. An initial pilot phase is due to be launched in early 2025.
YouTube also intends to develop a new technology that will enable celebrities from different sectors – be they creators, actors, musicians or athletes – to detect and manage AI-generated content featuring their face on the platform.
Finally, from the end of this year, YouTube should also enable creators to choose how third parties may or may not use their content.
Today, all content uploaded to YouTube over time contributes, in compliance with the conditions accepted by creators, to the machine learning and AI applications developed by Google. This includes improving its recommendation systems and developing new features such as automatic dubbing.
Faced with the onslaught of deepfakes on the Internet, other initiatives are taking shape, such as that of antivirus specialist McAfee, which recently announced the imminent arrival of its first deepfake detection tool for PC. Called Deepfake Detector, it will analyse videos in just a few seconds and alert users if it detects computer-generated content. It will be available exclusively on selected Lenovo PCs later this year, initially in the US, the UK and Australia. – AFP Relaxnews
