New findings suggest over 20% of Shorts shown to new YouTube users may be AI-generated

Your YouTube feed may be filling with videos made by nobody

The report highlights a surprising chunk of what new users see on YouTube Shorts is not exactly human-made.

In a fresh account test of 500 Shorts, Kapwing labeled 104 videos as AI slop and 165 as brainrot, which the researchers summarize as roughly 21 percent and 33 percent of that sample, respectively.

If your feed has started to feel uncanny, repetitive, or strangely contentless, this is the trend behind it.

Kapwing tried to recreate the experience of a fresh YouTube viewer

To avoid a personalized algorithm, researchers created a brand-new YouTube account and began scrolling Shorts like a first-time user. They tracked the first 500 recommendations and then labeled videos they believed were AI-generated or of poor quality, based on obvious automation signals.

In that sample, 104 videos were categorized as AI slop and 165 as brainrot, showing how quickly synthetic or low-quality content can appear.

AI slop is the internet’s new junk food category

Slop has become shorthand for digital content that is cheap to produce, easy to scale, and designed to harvest attention rather than inform or entertain. Merriam-Webster chose “slop” as its 2025 Word of the Year, defining it as low-quality digital content typically produced in large quantities using AI.

Brainrot is the wider neighborhood, covering slop plus other mindless formats that keep you swiping without remembering why.

The study argues that this is no longer a niche problem

Kapwing did not only watch a single feed. It also surveyed 15,000 of YouTube’s most popular channels, essentially the top 100 channels in each country, to gauge the size of the ecosystem. It identified 278 channels that exclusively upload AI-generated content.

Collectively, those channels were estimated to have gathered about 63 billion views and roughly 221 million subscribers, an enormous footprint for disposable content.

Recommendation algorithms reward volume and repetition

Here is the uncomfortable incentive loop. AI tools enable the generation of dozens of variations quickly, testing which one hooks viewers, and then scaling the winning formula.

Platforms like YouTube are giant A/B testing machines, so content that keeps people watching often gets amplified, even when it is shallow. Slop creators can iterate faster than humans, which means the algorithm sees constant fresh uploads and keeps sampling them.

The boom is global, and its impact varies by country

Kapwing’s report paints a surprisingly international map. In subscriber counts, it highlighted big audiences in places like Spain, Egypt, and the United States.

In terms of view counts, it highlighted a massive reach in countries such as South Korea and Pakistan, suggesting that the format travels easily across languages.

The common thread is not culture, it is mechanics: short, loopable clips that work even when you barely understand the context.

A few viral channels show why this content spreads so fast

Consider the extremes the report highlighted. India’s Bandar Apna Dost uses absurd AI characters and simple action beats that are easy to follow. Singapore-based Pouty Frenchie leans into kid-friendly fantasy scenes.

Kapwing points to a US-based Spanish-language channel called Cuentos Facinantes, which leans on anime-inspired storylines, including Dragon Ball-style tales.

In South Korea, Three Minutes Wisdom serves up short, highly repeatable clips of cute pets triumphing over wild animals, using different packaging but the same template-driven strategy that scales.

There is real money in slop, even when the quality is low

Kapwing estimates the broader slop ecosystem could generate around $117 million a year, with individual channels potentially pulling in millions annually. That does not always mean straightforward YouTube ad revenue, especially if repetitious uploads face limits.

Some creators mix in affiliate links, sponsorships, or off-platform funnels. Cheap production plus distribution can create a factory. When the cost of publishing is near zero, volume becomes the primary business model.

Kids may be the most vulnerable audience for this content to capture

Many of the examples in the report appear to be aimed at young viewers, featuring bright characters, simple plots, sound effects, and endless novelty. The risk is not just annoyance.

It is that kids can be nudged into compulsive viewing of content with little educational or social value, and parents may not realize how automated it is. If a feed is shaped by what holds attention, children become an obvious target market.

YouTube says the tool is neutral, but enforcement is messy

YouTube’s public stance is that generative AI is merely a tool that can produce both high- and low-quality content, and that all uploads must adhere to community guidelines.

The platform has tightened its rules around spam and inauthentic behavior, and has taken action against some fake channels, including high-profile AI-generated movie trailer pages. However, YouTube is also rolling out additional AI creation tools, which complicates the cleanup story.

Creators feel the squeeze

If you make videos the slow, human way, slop can feel like an unfair spam flood. It can crowd out recommendation slots, distort trends, and prompt creators to simplify or sensationalize their content to stay competitive.

At the same time, some legitimate creators are using AI responsibly for editing and localization, so the battle is not humans versus machines. It is authenticity versus automation at scale, and the algorithm is the referee.

You can train your own feed to dodge the worst of it

Small actions make a significant difference. Use the ‘Not interested’ control quickly, especially on the first unusual video, because early signals shape your feed. Subscribe to creators you trust and watch from the Subscriptions tab when you want a cleaner experience.

Be wary of channels with generic names, identical thumbnails, and nonstop uploads. When a clip feels like a loop with no point, treat it as a warning sign.

For a real-world example of how fragile these systems can be, it’s worth checking out one YouTuber’s claim that an AI moderation mistake wiped out his 350,000-subscriber channel.

The next phase will be a labeling and detection arms race

Expect increased pressure for more transparent disclosure, stronger detection, and stricter monetization rules for mass-produced content.

Slop makers will adapt because incentives are strong and tools continue to improve. So the realistic outcome is not a total purge; it is an ongoing tug-of-war. For viewers, the best defense is intentional.

And if you’re curious how platforms are still experimenting with incentives on the other side of that tug-of-war, take a quick look at YouTube TV’s new discounted pricing deal with Verizon.

What do you think about new claims that roughly one in five Shorts shown to new YouTube users may be AI-generated or low-effort content? Please share your thoughts and drop a comment.

This slideshow was made with AI assistance and human editing.

Don’t forget to follow us for more exclusive content on MSN.

Read More From This Brand:

• YouTubes CEO sets strict social media rules for his kids just like other tech leaders
• YouTube rolls out new Recap feature to show your yearly viewing habits
• YouTube TV or its rivals? The surprising differences between Hulu Live, Fubo, and Sling TV