Study finds YouTube algorithm pushing AI-generated ‘Junk' content to children

A new investigation finds that low-quality AI-generated videos are often recommended to children in the guise of educational content by the YouTube algorithm.

The study conducted by The New York Times reveals that the platform suggests junk “Shorts” from channels that apparently claim to teach toddlers basic education content like the alphabet, animals, and skills, etc.

The report published on Friday, February 27, analysed 1,000 Shorts that were recommended to young children.

A disturbing pattern was found that states that most of the suggested content is low-quality and AI-generated.

The videos showcase animals and people with unusual facial features, more than usual limbs and chaotic imagery.

These channels usually upload 30-second videos due to the limitations of AI tools. They significantly contain false information, yet they appear as educational.

The same pattern is also identified on YouTube Kids, a separate section designed by YouTube with additional parental controls.

Despite improved safety measures, researchers found that algorithm suggests similar content there as well.

Following the backlash, YouTube’s parent company, Google removed the channels from the YouTube Partner Program which enables creators to monetise their content. Various individual videos that violated the platform rules are also removed.