More than 20 per cent of videos recommended to new YouTube users consist of low-quality, AI-generated content, according to a new study by video-editing platform Kapwing.
The research analysed the top 15,000 YouTube channels globally, including the top 100 channels from each country, to assess the prevalence of automated, low-effort content in the platform’s recommendation system.
The study found that 278 channels exclusively posted AI-generated videos, collectively attracting over 63 billion views and 221 million subscribers. These channels are estimated to generate approximately $117 million in annual revenue.
To understand how frequently such content appears in recommendations, researchers created a new YouTube account and analysed the first 500 videos suggested by the platform. Of these, 104 videos — more than 20 per cent — were identified as AI-generated “slop”. Around one-third fell under a broader category referred to as “brainrot”, which includes repetitive or attention-driven low-quality content.
AI slop typically refers to videos created using automated tools with minimal human input, often relying on repetitive visuals, synthetic narration, or recycled material. Critics say such content prioritises engagement metrics over quality, creativity, or informational value.
The findings align with earlier reporting. A Guardian analysis earlier this year found that nearly 10 per cent of YouTube’s fastest-growing channels consisted largely of AI-generated content, despite the platform’s stated efforts to limit inauthentic or misleading material.
The study highlights growing concerns over how generative AI is reshaping digital content ecosystems. As AI tools become more accessible, researchers warn that algorithm-driven platforms may increasingly favour volume and engagement over originality and substance.
AI slop refers to low- to mid-quality content — including video, audio, text, or images — produced rapidly using AI tools, often with little attention to accuracy or originality. Its increasing presence is widely viewed as a challenge for maintaining content quality and trust online.