Report claims 60% of TikTok videos are AI slop, along with 21% on YouTube

The modern social media feed is rapidly transforming into a synthetic playground where human creators are becoming a minority. It’s no longer just a trend; it’s an algorithmic takeover that’s hard to ignore. A major industry report by video-editing platform Kapwing claims 60% of TikTok videos served to new users are low-quality AI slop, compared to 21% on YouTube.

This massive wave of machine-made uploads is changing how users interact with digital platforms. It’s a shift that’s leaving casual viewers feeling disconnected and increasingly frustrated with their daily scrolling. Social media feeds are slowly turning into a dull echo chamber of automated voiceovers and stolen clips.

The rise of automated junk feeds

A massive study by video editing platform Kapwing has exposed the true scale of this digital saturation. Researchers manually analyzed over 10,742 TikTok videos across 20 popular categories to assess the prevalence of synthetic content. To get an unbiased look, they also tracked the first 500 videos shown to a brand-new user.

The results of this raw feed experiment were incredibly surprising. Out of those first 500 recommended videos, a staggering 294 were identified as AI slop. That means nearly six out of ten videos shown to a new account are automated junk.

TikTok’s initial feed seems designed to push high-volume, low-effort content right from the start. The platform has been flooded with content, and it had already labeled 1.3 billion videos as AI-generated by late 2025. This automated deluge suggests that the platform’s initial charm is getting buried under machine-generated noise.

Breaking down the platform numbers

The study shows that TikTok’s AI problem is significantly worse than its competitors. While TikTok serves 59% AI slop to new accounts, YouTube Shorts serves a much more modest 21%. Essentially, TikTok serves nearly three times as much automated junk as YouTube.

However, YouTube is still facing its own battle with low-quality content. Around 33% of recommended YouTube Shorts fall into a broader category known as “brainrot.“ Many users report that their homepages are filled with soulless “Top 10” channels using AI narrators.

This contrast highlights the different algorithmic choices made by each platform’s engineering teams. TikTok seems to prioritize rapid, cheap engagement to keep viewers hooked. Meanwhile, YouTube’s recommendations rely on slightly more established signals before pushing automated channels.

Why the algorithm loves low effort

Creating a million mediocre TikToks with generative AI is much easier than creating one high-quality video. Platforms reward pure volume and cheap views, which heavily incentivize automated bot farms. As a result, low-effort creators are using automated tools to farm millions of views overnight.

This algorithmic design has led to severe imbalances across different content categories. Physical and highly visual categories are currently the safest from automation.

It’s incredibly easy for AI to stitch together stolen historical facts or generic science tips. Channels like “Chloe VS History” use realistic AI models to generate entire lessons in seconds. Unfortunately, these rapid uploads are often littered with embarrassing historical and factual errors.

Because AI models simply predict patterns, they often introduce biases and outright lies. In contrast, fitness and fashion creators still need a human body, which helps keep those niches clean.

The high cost for young minds

The absolute darkest corner of the AI slop problem is the content aimed directly at children. Over half of the videos in TikTok’s kids category are identified as AI-generated junk. Even worse, the hashtag #cartoonkids has reached an unbelievable 97% AI-slop rate.

Only three out of every one hundred videos under that kids’ tag are actually made by humans. Other popular tags, like #cartoons and #babysong, are also heavily saturated with 83% AI slop. This creates a massive pool of bizarre and nonsensical content for the youngest viewers.

These automated kids’ videos aren’t just low-quality; they’re sometimes completely nonsensical. An automated counting video under the #preschoollearning tag got the numbers completely wrong. The synthetic voices are often unsettling, and the comments sections are filled with robotic spam to boost views.

Pediatric experts are deeply worried about what this does to early brain development. Dr. Dana Suskind from the University of Chicago warns that this is toddler AI misinformation at scale. With millions of neural connections forming daily, these erratic videos are literally wiring developing brains in the wrong direction.

Unfortunately, parental oversight on these mobile screens remains critically low. Toddlers are frequently left alone to consume endless streams of automated garbage on these apps.

Fighting back against the machine

Frustrated users aren’t just sitting back and accepting this new synthetic reality. Comments sections are filled with angry viewers calling out obvious AI voiceovers and lazy scripts. Many viewers are now turning off auto-play and selecting “Don’t Recommend Channel” to block the slop.

The platform owners are feeling the heat and trying to make adjustments. TikTok scaled back its AI summaries in May 2026 after they made hilarious mistakes. They also rolled out tools allowing users to request less AI-generated content in their feeds.

However, many tech experts believe these passive settings aren’t nearly enough. While TikTok relies on user settings, YouTube has taken a much tougher, top-down approach. In early 2026, YouTube deleted 16 massive AI-slop channels with billions of views for inauthentic behavior. Regulatory pressures are also mounting against these tech giants. 

The bottom line on the synthetic wave

The era of passive scrolling is officially over, replaced by an ongoing battle for authenticity. With nearly 60% of new TikTok feeds covered in AI junk, the default internet has gotten incredibly messy. To escape the slop, users must actively train their algorithms by engaging only with genuine human creators.

Disclaimer – This list is solely the author’s opinion based on research and publicly available information. It is not intended to be professional advice.

Like our content? Be sure to follow us