Skip to content

YouTube unveils AI detection tools to safeguard content creators

  • by
  • 3 min read

YouTube announced two new AI detection tools to protect content creators from unauthorised use of their likeness. The measures primarily target artists, musicians, actors, athletes, and other public figures, focusing on misusing their voices, faces, and performances by AI-generated content.

These tools are set to expand YouTube’s well-established Content ID system, which already identifies copyright-protected material on its platform.

One of the major upgrades includes the introduction of ‘synthetic-singing identification technology,’ specifically designed to detect when AI is used to simulate someone’s singing voice.

“We’ve developed new synthetic-singing identification technology within Content ID that will allow partners to automatically detect and manage AI-generated content on YouTube that simulates their singing voices,” announced Amjad Hanif, Vice President of creator products, YouTube. “We’re refining this technology with our partners, with a pilot program planned for early next year.”

This could significantly impact the music industry, where concerns about AI mimicking artists without consent have grown louder. Additionally, YouTube is working on a system to detect when AI-generated videos feature the faces of individuals, aiming to prevent misleading or malicious use of someone’s image.

Beyond simply detecting AI-generated content, YouTube is also addressing a more complex issue: the unauthorised use of its platform’s content to train AI models.

Photo: Tada Images / Shutterstock.com
AI companies are scrapping YouTube videos to train AI models. | Photo: Tada Images / Shutterstock.com

This has been a flashpoint for creators who feel companies like OpenAI, Google, and Nvidia have trained their models using their work without permission or compensation.

While the company has yet to disclose specific details on how it plans to tackle this problem, YouTube hinted at new features that will allow creators greater control over how their parties use their content for AI training.

The platform promised more information later in the year but acknowledged that this issue requires a delicate balance between protecting creators and enabling the technological progress that AI offers.

Despite having previously promised content creators to compensate them for using their work in AI-generated music, YouTube has not yet addressed this issue. Last year, YouTube began working with Universal Music Group (UMG) to find a solution, signalling that its Content ID system would be at the forefront of identifying when and how AI systems are using music.

However, YouTube reassured creators that a pilot program focusing on synthetic singing detection will begin early next year.

In the News: Critical flaw in LiteSpeed Cache plugin puts 6 million sites at risk

Kumar Hemant

Kumar Hemant

Deputy Editor at Candid.Technology. Hemant writes at the intersection of tech and culture and has a keen interest in science, social issues and international relations. You can contact him here: kumarhemant@pm.me

>