According to Mashable, a new study from video-editing company Kapwing, reported by the Guardian, has quantified the AI slop problem on YouTube. The research found that more than one in every five videos (21%) shown to new users by the YouTube Shorts algorithm is low-quality, AI-generated content. In a test of the first 500 videos on a brand-new Shorts feed, 104 were AI-generated and another 165 were classified as “brainrot,” totaling over half of the content. The study also revealed the geographic scale of the issue, noting that AI slop channels in Spain have a combined 20.22 million subscribers, while the U.S. has nine such channels in its top 100 and 14.47 million slop subscribers. The report makes it clear this AI-generated content is proliferating across social media feeds, from fake animal clips to other bizarre, algorithmically-generated videos.
The Algorithmic Pipeline
Here’s the thing about this study’s method: it’s revealing because it used a fresh, untouched YouTube account. That’s basically looking at what the platform’s default, cold-start algorithm decides is a good first impression. And its first impression is that 21% AI slop is acceptable. That’s a huge signal about the raw material flooding YouTube’s upload pipelines. The system is clearly being gamed by channels that can pump out volumes of this stuff to chase trends and clicks. And when the algorithm has no prior viewing history to go on, this cheap, engaging, often bizarre content is what it surfaces. It makes you wonder, what does YouTube’s own internal metrics say about the watch time and engagement on these videos compared to human-made content? If it’s high, the problem is only going to get worse.
A Global Slop Economy
The country-by-country data is fascinating, too. Spain leading in total subscribers to these channels is one thing. But the U.S. having nine AI slop channels in its top 100 is arguably more alarming, given the size and influence of that market. It points to a kind of global, low-tier content economy that’s emerged. Creating these videos is incredibly cheap and scalable—you need an idea, some AI image/video generation tools, a text-to-speech engine, and maybe some stock music. The barrier to entry is almost zero. So, for creators in any country, it becomes a viable, if soulless, path to potentially massive algorithmic reach and ad revenue. The subscriber numbers in the millions prove there’s an audience, or at least an algorithm, that’s consuming it.
Is There Any Way Back?
So what’s the endgame? Platforms like YouTube are in a tough spot. They’ve built empires on the promise of infinite, free content. AI is now delivering “infinite content” in the most literal, depressing sense possible. But cracking down too hard is messy. How do you consistently and fairly define “low-quality AI slop” at a scale of billions of videos? Do you penalize a genuinely creative use of AI tools? The line is blurry. Meanwhile, for users, it creates a weird erosion of trust. When you know a significant chunk of what you’re seeing is synthetic noise designed purely for retention, it changes how you interact with the platform. You start to feel like you’re not just watching videos, you’re being processed by a content farm. The Kapwing study and the Guardian’s report aren’t just highlighting a content problem—they’re pointing to a fundamental credibility crisis for the biggest video platform on earth. And that’s much harder to fix than tweaking an algorithm.
