How Korea Became the World's #1 Producer of 'AI Slop'

Scroll long enough and you start to feel it: more and more videos that are off somehow. Smooth on the surface, but the captions read strangely and nothing actually stays with you. There’s now a name for this stuff — AI slop. And here’s the surprise: the country that produces the most-watched AI slop in the world isn’t the US. It’s Korea.

That makes Korea an unusually clear window into where the whole internet is heading. This piece walks through what slop actually is, why Korea ended up #1, and how to tell the real thing from the flood pouring into your feed.

Key takeaways

• “AI slop” means low-quality content mass-produced with generative AI, and “slop” was named 2025’s word of the year — Merriam-Webster / American Dialect Society.
• Korea-based AI-slop channels logged about 8.45 billion views — the most in the world, far ahead of #2 Pakistan (~5.3B) and #3 the US (~3.4B) — Kapwing.
• YouTube cracked down, deleting 16 mass-production channels with a combined 35 million subscribers — Herald Economy.
• The real line isn’t “did they use AI” but “is there human creative involvement.” Use AI and still add human judgment and editing, and it isn’t slop.

What AI slop actually is

AI slop is content churned out fast and cheap by generative AI, lacking effort, quality, or meaning. The word “slop” originally means the food scraps or swill fed to livestock — lots of volume, no nutrition. The term jumped straight into the digital world to describe exactly that.

Its status was sealed in 2025: Merriam-Webster named “slop” its word of the year, and the American Dialect Society chose the same word. When the word that best explains a year is “low-quality AI content,” it means a lot of people felt — and disliked — the phenomenon.

The crucial point: slop does not simply mean “content made by AI.” Content where a human plans it, checks the facts, and refines it is not slop. The defining trait of slop is mass production without human judgment — cranking out the same template, chasing views and ad revenue, with no verification and no editing.

META TOUR’s take: the real issue in the slop debate isn’t “AI vs human.” It’s whether there’s any intent to make meaning. A hammer isn’t a bad tool, and neither is AI. The problem isn’t the tool — it’s stamping out volume with no thought behind it. Miss that distinction and you slide into the wrong conclusion that “AI is always bad” — when in reality people are increasingly turning to AI for comfort, not just spam.

Why did Korea become #1?

Start with the most striking data. According to an AI-slop report from the video platform Kapwing, Korea-based AI-slop channels racked up a cumulative 8.45 billion views — the most of any country surveyed — Kapwing. That’s well ahead of #2 Pakistan (~5.3B) and #3 the US (~3.4B). Five of the top 10 channels by views were Korean.

One clarification matters: this data means “Korean-language slop channels logged the highest views,” more than “Koreans love slop the most.” Still, being #1 isn’t something to wave away.

So why Korea? It’s hard to be definitive, but a few factors come up together. First, Korea is a market with extremely concentrated short-form video consumption — the habit of swiping rapidly through short clips is fertile ground for slop. Second, in an economy where views convert directly into ad revenue, AI mass-production — which costs almost nothing to make — becomes a very high-margin business. Third, AI video tools dramatically lowered the language barrier, so anyone can now stamp out Korean-language content at scale.

Korea’s slop spans genres, too: AI-composited animal videos, meaningless “brainrot” clips, and fabricated “national-pride” story channels claiming some foreign celebrity praised Korea — all mass-produced. The problem is that as these multiply, genuinely careful content gets buried.

How did there get to be so much of it — the zero-cost trap

The explosion of slop is explained by simple economics. The key: the cost of production has collapsed toward zero. Making a single video used to take real time and money — planning, shooting, editing, voice recording — so you needed a certain number of views just to break even. Generative AI broke that formula. A chatbot writes the script, an image model makes the visuals, a synthetic voice does the narration. The human just presses a button.

Now the math flips entirely. When production costs near zero, a video that gets even a few hundred views can turn an ad or affiliate profit. Make 100 videos, have 99 flop, and if just one hits, you’re ahead. So “make a lot” becomes a more rational strategy than “make it good.” Slop is less the product of laziness than the result of a perfectly rational chase for revenue — the same logic that drives AI selling emotional “certainty” to shoppers.

META TOUR’s analysis: blaming the people who make slop doesn’t solve it. As long as mass production pays, the makers keep coming. That’s why platforms are reworking monetization and why search and recommendation algorithms are moving to filter out mass-produced content. Slop shrinks only when “slop doesn’t pay” becomes true.

It’s not just YouTube — slop is spreading everywhere

Slop doesn’t stay in short video. Wherever the “zero production cost” logic applies, it spreads.

Music is a prime example. According to the streaming service Deezer, as of April 2026 about 44% of newly uploaded tracks were AI-generated, roughly 75,000 songs a day. More striking: up to 85% of streams of fully AI tracks were fraudulent (bots inflating the play counts) — Deezer. In summer 2025, an “AI band” (The Velvet Sundown) gathered around a million monthly Spotify listeners while hiding its nature — Euronews.

News isn’t safe either. The misinformation watchdog NewsGuard tracks 3,006 AI-generated “news” sites operating without human oversight across 16 languages, Korean included — NewsGuard. And the bigger picture is heavier still: security firm Imperva found that in 2025, for the first time, automated bot traffic topped 51% of all web traffic — passing humans — and rose to 53% in 2026 — Imperva. An internet where machines talk more than people. Slop is just one slice of it.

The decisive difference between slop and real content

Explanation only goes so far, so let’s compare directly. Say two creators tackle the same topic — a self-help staple: “the morning habits of successful people.” (The examples below illustrate the difference; an AI’s actual output varies each time, even for the same prompt.)

The difference is clear. Same topic, same AI, and yet the results are in different classes. What made the gap? Four things:

• Fact-checking — the slop version asserts “wake early = success” with no verification. The done-right version doubts the cliché and checks real evidence.
• Perspective — the slop version is an obvious list anyone knows. The done-right version offers its own angle: consistency over wake-up time.
• Honesty — the slop version overpromises (“do this and you’ll succeed”). The done-right version states the limits and the reality.
• Human editing — the slop version ships the raw output. The done-right version chooses and refines the sentences so they reach a reader.

META TOUR’s take: the interesting part is that the done-right version can also use AI. What makes the difference isn’t “was AI used” but “did a human ship the raw output, or verify and refine it.” That process is exactly the line between slop and content.

Why YouTube and Google drew their knives

Platforms aren’t standing by. The most aggressive is YouTube. CEO Neal Mohan said YouTube filters low-quality AI content through its spam and clickbait systems to curb its spread — Herald Economy. The targets: content that’s mass-produced, repeats an identical template, or is cranked out mechanically with no human creative involvement.

The blade was sharp. The 16 AI mass-production channels YouTube removed had a combined 35 million subscribers, 4.7 billion cumulative views, and an estimated annual revenue in the tens of millions of dollars — Herald Economy. Among them was a Spanish-language channel with 6 million subscribers. Not a mere warning — entire channels and their revenue vanished.

The most important message: YouTube did not ban “using AI.” Creators who use AI tools are still welcome. The blade aimed at mass-produced content with zero human creative involvement — abnormally high upload frequency, identical format across every video, no original commentary, no trace of editing. Pile up those signals and a channel gets flagged.

Search follows a similar line. Google already names “scaled content abuse” in policy and has said it will filter mass-produced low-quality content meant to manipulate rankings — Google. The test isn’t whether AI was used, but whether it was made to help a searcher or churned out just to chase rank.

The area of greatest concern is children’s content. Experts note that only around 5% of videos labeled “educational” are actually high quality. In response, more than 230 people — 135 organizations and 102 experts — signed a letter demanding YouTube block AI slop — Fortune. The less developed a child’s judgment, the more easily they’re exposed to meaningless mass-produced content.

How to spot AI slop

Platform enforcement aside, protecting your own feed comes down to you. The good news: slop is fairly easy to catch from a few signals.

• Off details — unnatural intonation in a synthetic voice, wrong captions, subtly warped images or fingers. The faster it’s made, the more the details give it away.
• Meaningless repetition — padding the same point in different words, or circling generalities with no conclusion. You finish and nothing remains.
• No source or author — confident claims poured out with no evidence or source, and no clear sense of who made it.
• Abnormal upload frequency — one channel posting dozens a day, all in the same format, is likely mass production.

When you see these signals, the response is simple: check the channel’s past videos and history, doubt unsourced assertions, and use “not interested” or report functions. Small actions, but they tell the recommendation algorithm, “I don’t want this.”

What the slop era leaves us — three meanings

AI slop isn’t a story you can end with “there’s more junk on the internet now.” It carries at least three meanings.

First, the standard of value for content is shifting. The age of competing on volume is fading. Now that anyone can produce infinitely, what’s become scarce is content with human judgment in it — fact-checking, an original perspective, honest editing. These become the bigger differentiators going forward.

Second, law and policy are catching up. Korea’s AI Framework Act, effective January 2026, requires that generative-AI output be labeled as AI-generated — with especially clear labeling for video, audio, and images hard to tell from reality (Article 31). The EU’s AI Act sets a similar disclosure duty under Article 50, with rollout in the second half of 2026 — European Commission.

Third, the consumer’s discernment is the real line of defense. Platform enforcement has limits, and mass-production tech keeps evolving. In the end, our choice to spend less time on junk and linger on careful work is what changes the ecosystem. An internet where slop doesn’t pay is built when people start telling slop apart.

The takeaway

That “slop” became the word of the year shows what kind of era we’re passing through. As the cost of making content collapses to zero, the internet fills fast with mass-produced junk — and Korea sits right in the middle of it. Being #1 in views isn’t a boast; it’s closer to a signal that Korea is facing this problem first.

But it’s not all grim. Platforms are reworking policy, laws are creating labeling duties, and people are sharpening the eye to tell slop apart. Most important: the core of the problem isn’t “AI” but “mass production without human judgment.” Next time you meet an off video in your feed, pause and check the signals — is there a source, a perspective, a human touch? Gather enough of that small discernment, and you’re a step closer to an internet where slop doesn’t pay. In an age drowning in volume, what’s truly become scarce is human care.

Sources

• Merriam-Webster / American Dialect Society — 2025 Word of the Year: “slop.” 2025-12. View source
• Kapwing — “The Global Rise of Low-Quality AI Videos” (Korea AI-slop channels 8.45B views, #1 worldwide). 2025. View source
• Herald Economy — YouTube deletes 16 large AI revenue channels (combined 35M subscribers, 4.7B views). 2026-02. View source
• Fortune — 135 organizations and 102 experts sign letter urging YouTube to block AI slop. 2026-04. View source
• Google — Search spam policies: scaled content abuse. View source
• Korea Policy Briefing — AI Framework Act in force; AI-output labeling duty (Article 31). 2026-01. View source
• Deezer Newsroom — AI-generated tracks 44% of new uploads, ~75k/day, up to 85% of full-AI streams fraudulent. 2026-04. View source
• Euronews — The Velvet Sundown: AI band, ~1M monthly Spotify listeners controversy. 2025-07. View source
• NewsGuard — AI Tracking Center: 3,006 AI content-farm news sites across 16 languages (Korean included). 2026-03. View source
• Imperva — Bad Bot Report: bot traffic 51% of the web (2025) → 53% (2026), passing humans. 2025–2026. View source