I did not expect a spreadsheet to make me uneasy. Then I sat down with the numbers behind one of YouTube's strangest income streams: videos made with AI, at huge volume, with almost no human craft behind them.
You have probably felt it. You open YouTube to watch one specific video, and a few recommendations later you land on a dog with too many teeth, or a cartoon monkey flying a tomato-shaped helicopter. You scroll past it. Most people do.
Researchers at the video-editing company Kapwing did not scroll past. They studied 15,000 of the most popular YouTube channels and estimated what this content earns: about $117 million a year. That is an estimate, not an audited number, but it rests on real channel data. Below are 20 sourced facts, plus practical ways to protect your own feed.
- The $117 Million Estimate Behind the Headline
- Inside the Kapwing Study: 15,000 Channels
- What a Brand-New Account Sees?
- 278 Channels Built Only on AI Video
- Bandar Apna Dost: The Monkey Channel
- Cuentos Fascinantes: The Most Subscribed Channel
- Pouty Frenchie: Candy Worlds for Young Viewers
- The AI World: Disaster Scenes as Sleep Content
- Spain and South Korea: The Global Spread
- Who Makes This Content, and Why?
- Why the Money Flows This Way?
- What YouTube's Policy Actually Says?
- Workslop: The Same Problem at the Office
- Platform Purges: What They Fix and What They Don't
- The Illusory Truth Effect
- Human Creators vs. a Machine That Never Sleeps
- The Disney and Sora Deal That Fell Apart
- Regulators and the EU AI Act
- What Platforms Say vs. What Incentives Reward?
- How to Spot AI Slop and Where This Goes Next?
1. The $117 Million Estimate Behind the Headline
The figure comes from a Kapwing AI slop report that asked a simple question: how much of what YouTube shows to new viewers is low-quality AI content, and how much money might it earn? Kapwing defines AI slop as careless, low-quality content made with automated tools and spread to farm views and subscriptions.
The estimate covers only the channels the researchers could identify and measure. Kapwing builds it from public channel data, so treat $117 million as a reasonable range rather than an exact payout.
2. Inside the Kapwing Study: 15,000 Channels
The researchers did not rely on their own feeds. As The Guardian reported, they surveyed the top 100 trending YouTube channels in every country, 15,000 in total. Of those, 278 contained only AI slop.
That is a global sample, not a single-language or single-market snapshot. It shows how far this content travels across borders.
3. What a Brand-New Account Sees?
To see what the algorithm does with no history, the team created a new YouTube account. Of the first 500 Shorts recommended to it, 104 (about 21%) were AI slop. Another 165 (about 33%) fell into the wider "brainrot" bucket, which includes AI slop and other low-effort, attention-farming content.
Some limits matter here. It was one test account, it covered the Shorts feed only, and results will vary by country and time. Still, it is a useful snapshot of the default experience for someone with no preferences set.
"Want the Shorts numbers in more detail? Read our YouTube Shorts AI slop data breakdown."
4. 278 Channels Built Only on AI Video
Inside the dataset, the 278 pure-slop channels together recorded more than 63 billion views and about 221 million subscribers. Kapwing estimates they earn roughly $117 million a year between them.
Subscriber counts overlap, so this is not 221 million unique viewers. Even so, it shows a parallel media ecosystem running beside the creator economy most people follow.
5. Bandar Apna Dost: The Monkey Channel
India's Bandar Apna Dost is the channel most often used to illustrate the trend. It features a hyper-realistic CGI monkey and a Hulk-style character traveling in a tomato-shaped helicopter.
TechSpot's coverage puts it at about 2.07 billion views, the most of any AI channel in the study, while other outlets cite figures closer to 2.4 billion because they measured at different times. Kapwing estimates it could earn up to $4.25 million a year.
6. Cuentos Fascinantes: The Most Subscribed Channel
Cuentos Fascinantes is based in the U.S. and publishes in Spanish. With 5.95 million subscribers, it is the most-subscribed AI slop channel in the study. Kapwing describes its Dragon Ball-themed videos as low quality, with about 1.28 billion views.
Because the videos borrow from an existing franchise, they also raise copyright questions that platforms and rights holders are still working through.
7. Pouty Frenchie: Candy Worlds for Young Viewers
Pouty Frenchie, based in Singapore, centers on an AI-generated French bulldog wandering through candy-colored fantasy worlds. Kapwing estimates nearly $4 million a year from roughly 2 billion views, and reports describe it as aimed at very young children.
Young viewers cannot tell AI-made videos from anything else on screen, which is why child-focused AI content gets extra scrutiny from parents and platforms.
"Parents often ask what YouTube is doing about this. See YouTube's AI slop kids ban explained."
8. The AI World: Disaster Scenes as Sleep Content
Not every channel in the dataset is harmless. Pakistan's The AI World publishes AI-generated short films about floods, with titles like "Poor Family" and "Kitchen in Flood," often over soothing rain-and-thunder sleep audio. The channel reportedly has about 1.3 billion views.
The concern is ethical rather than legal. Imagery of real-world disasters is being turned into background content for ad revenue, and viewers may mistake it for real footage.
9. Spain and South Korea: The Global Spread
Two countries stand out in Kapwing's data. Spain leads in subscribers, with about 20.22 million subscriptions to trending AI slop channels. South Korea leads in views: its 11 trending slop channels have passed 8.4 billion views, the highest of any country.
Subscriptions are not unique people, so this does not mean half of Spain follows these channels. The point is that content with no cultural or language barrier travels easily. A bulldog in a candy forest needs no subtitles.
"YouTube is not the only platform affected. Read how Facebook's AI slop push is spreading."
10. Who Makes This Content, and Why?
Much of the industry is driven by creators in middle-income countries such as India, Nigeria and Kenya, where YouTube earnings can exceed local median wages. For many of them this is a real income, not a joke.
That makes it hard to treat the individual creator as the villain. Most are responding rationally to what the system rewards. The real question is why the system rewards it.
"Creators using AI should know the warning signs. Check these signs your YouTube channel is demonetized."
11. Why the Money Flows This Way?
How Shorts revenue is split?
Most slop lives in Shorts. Long-form creators keep 55% of ad revenue on their watch pages. For Shorts, YouTube announced a pooled model in which ad revenue is combined, then split by each creator's share of Shorts views, and creators keep 45% of what is allocated to them.
A rough calculation
Divide Kapwing's $117 million by 63 billion views and you get about $1.86 per 1,000 views. If all of that came through the 45% Shorts share, the implied gross ad revenue would be about $260 million, with YouTube retaining roughly $143 million. This is our own back-of-envelope estimate, not a reported figure, and real revenue depends on the mix of Shorts and long-form.
The takeaway is simple. Platforms earn on every view, whether the video took three weeks or twelve minutes to make.
12. What YouTube's Policy Actually Says?
YouTube says generative AI is a tool that can produce both high- and low-quality content, and that it focuses on connecting users with high-quality videos however they are made. AI use alone is not a violation.
However, YouTube's inauthentic content policy makes mass-produced or repetitive content ineligible for monetization. It explicitly covers AI-generated content made from generic templates without the creator's original insight. YouTube says this has always been a requirement and renamed the policy in July 2025 for clarity. That means Kapwing's revenue figures describe what channels could earn, and some of these channels may be limited or demonetized.
"See how enforcement has played out in YouTube's AI slop purge."
13. Workslop: The Same Problem at the Office
Slop is not only a YouTube issue. BetterUp Labs and the Stanford Social Media Lab surveyed 1,150 U.S. desk workers and coined the term "workslop" for AI output that looks finished but lacks substance.
About 40% of respondents had received workslop in the past month. The researchers estimated a cost of $186 per month per affected employee, or roughly $9 million a year for a 10,000-person organization.
The pattern matches YouTube's. Volume gets rewarded first, the producer faces little downside, and someone else pays the cleanup cost later.
14. Platform Purges: What They Fix and What They Don't
Spotify's spam removals
Spotify said it removed more than 75 million spammy tracks in 12 months. Its AI protections announcement also introduced an impersonation policy, a spam filter and support for AI disclosures in music credits through the DDEX standard.
In a pro-rata royalty pool, removing spam mostly redistributes existing money rather than creating new money. That is our reading, and it explains why a purge helps quality without changing the underlying incentive.
"For the full music story, read our Spotify AI music purge report."
Channel deletions on YouTube
Removing channels at scale is also real action. It cleans up feeds, but it rarely changes what earns money.
"We covered one wave of takedowns in Google's deletion of AI YouTube channels."
15. The Illusory Truth Effect
Psychologists have long documented the illusory truth effect: the more often you see a claim, the truer it feels, even when it is false. Slop does not need to be convincing to exploit this. It only needs to be repeated.
Researchers worry about the cumulative effect. Scroll past enough synthetic faces, disaster shorts and uncanny animals, and your instinct to check what is real gets worn down.
16. Human Creators vs. a Machine That Never Sleeps
A slop channel can publish several videos a day, test dozens of thumbnails and adjust to whatever the algorithm rewards in near real time. A human creator making one researched video a week cannot match that volume, however good the work is.
Critics describe large platforms as A/B testing machines that reward what performs, not what is original. That is a structural gap, and no platform has announced a fix for it.
17. The Disney and Sora Deal That Fell Apart
In December 2025, Disney announced a three-year licensing deal with OpenAI, including a $1 billion equity investment and more than 200 characters for the Sora video app. It looked like proof that prompt-based video was going mainstream.
Then, on March 24, 2026, OpenAI said it was discontinuing Sora and winding down the Disney partnership, according to Bloomberg Law. The lesson is useful: even the biggest names are still testing what AI video is worth, and no business model is settled.
18. Regulators and the EU AI Act
Rules are catching up. Article 50 of the EU AI Act, which covers transparency for AI-generated content and deepfakes, has applied since August 2, 2026. The European Commission also finalized a Code of Practice on marking and labelling AI-generated content in June 2026, and generative systems already on the market have until December 2, 2026 to meet the marking requirement.
In music, the DDEX standard lets labels and distributors disclose how AI was used in a track's credits. In the United States there is still no single federal rule requiring platforms to label AI video, though some states are moving on their own.
19. What Platforms Say vs. What Incentives Reward?
The public message
Platforms say they take the issue seriously, are improving detection and remove content that breaks their rules. All of that is generally true.
The incentive underneath
- Recommendation systems optimize for watch time and engagement, and slop can perform well on both.
- Removals clean up feeds but do not change who gets paid for what.
- "AI is just a tool" is accurate, but the tool changes the cost of making content, and payouts and discovery have not adjusted.
None of this requires assuming bad faith. These companies are optimizing the metrics they always have, and slop is good at hitting them. Fixing it means changing what gets rewarded, not only what gets removed.
20. How to Spot AI Slop and Where This Goes Next?
A quick checklist
- Emotionally loaded but generic titles such as "Poor Family" or "You Won't Believe"
- Faces or animals that move almost naturally, with small lighting or motion glitches
- Calm music that does not match the subject, even for hardship content
- Several uploads a day from one channel, often on unrelated topics
- Generic, repetitive or disabled comments
A polished thumbnail proves nothing either way, since plenty of slop channels have professional ones and plenty of small human creators do not. The most useful habit is pausing before the next video autoplays. Slop depends on momentum.
"More red flags are covered in signs you are looking at AI slop."
Where this goes next?
My view after going through the data is that AI slop is not a glitch waiting for a patch. It is an economic system with its own communities, incentives and platform-level economics. The $117 million estimate is likely a floor, not a ceiling, as tools improve.
For viewers, the burden still falls mostly on attention and skepticism. For creators, the field is uneven. And regulation, while moving, is unlikely to change what shows up in your feed tonight.
Frequently Asked Questions (FAQ)
What exactly is AI slop?
AI slop is low-effort, mass-produced AI-generated video or imagery made to maximize views and ad revenue rather than to inform or entertain.
How much money does the AI slop economy make?
Kapwing estimates about $117 million a year across 278 channels that publish only AI-generated content, based on a survey of 15,000 popular YouTube channels. It is an estimate, not a confirmed payout.
Is AI slop against YouTube's rules?
Not by itself. YouTube allows AI-assisted content, but mass-produced or repetitive content, including AI content made from generic templates, is ineligible for monetization under its inauthentic content policy.
Which countries are most affected?
In Kapwing's data, Spain leads in subscriptions to trending AI slop channels, with over 20 million, and South Korea leads in views, with more than 8 billion.
Can I tell if a video is AI-generated?
Often, yes. Look for unnatural motion, generic titles, mismatched music and very high upload frequency, as listed in the checklist above.
Related Reading