Study after study now shows YouTube isn’t just recommending human-made clips anymore. A deep dive by video-editing startup Kapwing found that over 20% of the videos YouTube’s algorithm serves to brand-new users are what researchers call “AI slop” — low-quality, AI-generated videos that are churned out to grab attention and rack up views rather than to inform or entertain thoughtfully.
Kapwing’s team looked at the top 15,000 YouTube channels across the globe (the top 100 from each country) and discovered 278 channels that upload nothing but AI-generated content. Together, these channels have accumulated about 63 billion views and roughly 221 million subscribers, and industry estimates suggest they generate around $117 million in revenue each year.
To see how this plays out in practice, researchers created a fresh YouTube account and tracked the first 500 recommended videos. They found 104 of those were pure AI slop, while if you include other low-quality, attention-grabbing clips — a category dubbed “brainrot” — roughly one-third of all recommendations fell into this broader low-value bucket. That’s a massive chunk of initial impressions for someone just starting to explore content on the platform.
So what does “AI slop” actually look like? These are typically videos with AI-crafted visuals, voices, and scripts that don’t involve real production effort. They often lean into bizarre scenarios, absurd characters, or sensational hooks designed to pull in viewers fast. Examples highlighted in the study include channels featuring anthropomorphic animals in surreal battles or pets in hyper-edited adventures.
Part of why this content proliferates is economic: creating AI slop takes far less time and money than traditional video production. People with minimal resources can generate thousands of videos using generative tools, tailor them for virality, and capitalize on YouTube’s ad system or other monetization strategies. Even if individual videos are low quality, the volume and clickbait-style titles and thumbnails help them rack up views.
YouTube says it remains committed to promoting high-quality, policy-compliant content regardless of how it’s made, but critics argue the platform’s recommendation engine still heavily favors engagement signals over depth or originality. That means a new user might see a feed dominated by these cheap, mass-produced clips before YouTube’s algorithm starts learning their real preferences.
This trend raises broader questions about the future of online video content and the role of AI in shaping what users see. There’s a growing “shadow economy” of creators who exploit AI tools to flood social platforms with endless low-effort clips, outpacing many human creators in sheer output. As these AI channels grow, they could reshape how audiences perceive quality, influence ad markets, and even affect the kinds of creators who can succeed on platforms like YouTube.
At the same time, there’s pushback from users and creators who feel the glut of AI slop dilutes the platform’s value. Some argue for better filters or transparency about AI-generated content, while others worry that real human creativity will be drowned out by algorithmically optimized noise if platforms don’t adjust how they rank and recommend videos.
In short, YouTube in 2025 isn’t just a place where humans share stories, tutorials, and vlogs anymore — it’s also a battleground where AI-generated content tries to hijack attention, revenue, and recommendation space, especially for new users whose feed preferences haven’t been shaped yet.