YouTube's algorithm is serving AI slop to new users
A new study from Kapwing found that over 20% of the videos YouTube recommends to new accounts are low-quality "AI slop." Researchers created a fresh profile and discovered that 104 of the first 500 suggestions were automated junk, estimating these copycat channels generate around $117 million annually. While platforms scramble to moderate, low-effort synthetic media is already dominating feed recommendations at scale.
YouTube’s algorithm has a junk food problem, and it's getting worse.
A new study from video-editing platform Kapwing reveals that over 20% of the videos recommended to brand-new YouTube accounts are "AI slop" — low-effort, auto-generated content explicitly designed to farm views.
Researchers spun up a fresh account and examined the first 500 clips pushed to its feed. A staggering 104 videos fit the AI slop description, while roughly another third fell under the general umbrella of "brain rot."
Monkeys, demons, and millions in cash
Kapwing dug into 15,000 of the platform's most popular channels to see just how deep this rabbit hole goes. They identified 278 channels that post nothing but AI slop, collectively racking up 63 billion views and 221 million subscribers.
Even though YouTube’s official guidelines technically restrict low-quality, mass-produced synthetic media from monetizing, the cash is clearly flowing. Kapwing estimates these 278 channels pull in around $117 million in revenue annually.
The undisputed heavyweight is an Indian channel named Bandar Apna Dost. Its feed features an anthropomorphic rhesus monkey teaming up with a Hulk-style muscular creature to battle demons. That channel alone has cleared over 2.4 billion views, pulling in an estimated $4.25 million a year.
Viewership is heavily international. South Korea led all regions with 8.45 billion slop views, followed by Pakistan at 5.34 billion and the U.S. at 3.39 billion, while Spain-based channels netted the most subscribers.
Big Tech’s double standard
YouTube insists it is keeping a close watch. A spokesperson told The Guardian that generative AI is just a tool capable of producing both high- and low-quality output, emphasizing that policy-violating videos get pulled. Just earlier this month, YouTube terminated two massive operations pushing fake AI movie trailers.
Yet the platforms are also fueling the very fire they claim to fight.
YouTube recently baked Google’s Veo 3 video generator straight into Shorts, encouraging users to generate clips on the fly. Meanwhile, Meta CEO Mark Zuckerberg openly told investors in October that AI-generated media will represent a massive new corpus of content for recommendation feeds across Facebook and Instagram.
Why it matters
The "Dead Internet Theory" is quietly turning into everyday reality. When one in every five videos fed to a clean account is synthetic sludge, the web's signal-to-noise ratio takes a massive hit.
What makes this trend so stubborn is simple economics: it works. Real human creators spend days scripting and editing, only to compete against automated pipelines churning out talking animals fighting monsters for virtually zero marginal cost. The fact that these channels pull billions of views means a massive segment of the audience either cannot tell the difference or simply doesn't care.
We think YouTube and its peers are playing a hypocritical game here. Publicly, they posture as guardians of quality; internally, engagement is engagement, regardless of whether a human or a prompt produced it. Until platforms actively penalize synthetic sludge in the recommendation engine instead of merely policing the worst copyright offenders, human creativity will keep getting crowded out by the slop machine.
Sources (1)
- indianexpress.com(opens in a new tab)Originalindianexpress.com