I spend a lot of my week auditing AI content across social platforms for this blog, and even I wasn't ready for what one French college student uncovered on Facebook. What started as one guy screenshotting a bizarre photo for laughs turned into a 133,000-follower watchdog account, a term the dictionary officially recognized, and a paper trail leading straight to a $300,000 payday. Here's the full story, the numbers behind it, and exactly how to spot this stuff in your own feed.
- The Image That Broke Everything
- Meet Théodore Cazals: The Student Who Took on Facebook
- What Exactly Is "AI Slop"?
- The Shocking Numbers: How Bad Is It Really?
- The $300,000 Question: Who's Making Money From This?
- How AI Slop Actually Works
- Meta's Response: Too Little, Too Late?
- Why This Matters for You
- How to Spot AI Slop
- The Future of Social Media
- Frequently Asked Questions (FAQ)
The Image That Broke Everything
The Photo That Made a Student Snap
Picture two emaciated children sitting in the pouring rain on a busy road. They have thick beards despite their boyish faces. One has no hands and only one foot. The other holds a sign saying it's his birthday, asking for likes. A birthday cake sits inexplicably beside them in the middle of traffic.
This image, riddled with obvious AI artifacts, went viral on Facebook. It pulled in nearly one million likes and heart emojis.
For Théodore Cazals, a 20-year-old student from Paris, something finally snapped. Describing the moment later, he said the flood of absurd AI images racking up engagement with zero scrutiny felt, in his own words, simply "insane to me."
The Birth of a Rebellion
Théodore didn't just get angry. He got active.
He created an account on X (formerly Twitter) called "Insane AI Slop" (@FacebookAIslop) and started calling out the fake content, mocking it, and exposing it for what it was.
Within months, his following exploded to over 133,000 people.
His inbox became flooded with submissions from around the world. Common themes emerged quickly:
- ✅ Religious imagery (think "Shrimp Jesus," AI-generated images of Jesus fused with crustaceans)
- ✅ Military propaganda
- ✅ Poor children doing "heartwarming" things
- ✅ Elderly people celebrating impossible birthdays
- ✅ Cats in dramatic soap opera scenarios
Why This Story Matters Right Now?
Théodore's exposé isn't just another viral moment. It's a warning sign for what's happening across social media.
Facebook in 2026 is not your grandmother's Facebook.
Mark Zuckerberg himself has described social media as entering a new, third phase, one where AI-generated and AI-remixed content sits alongside friend posts and creator content as a major new pillar of the feed.
The problem? That new corpus of content is increasingly fake, low-quality, and designed to manipulate, not inform or connect.
You might also want to read: Meta's own admission that Facebook has entered its AI content "phase 3"
Meet Théodore Cazals: The Student Who Took on Facebook
From Paris to Global Attention
Théodore Cazals isn't a tech journalist. He's not a cybersecurity expert or a disinformation researcher.
He's just a college student who got fed up with what he was seeing on his screen.
The image of those two AI-generated children with beards, the one that got nearly a million likes, was the breaking point, as he later described in an interview about the growing AI slop backlash.
What He Discovered?
As Théodore dug deeper, he uncovered patterns that would shock even the most cynical internet user.
Common AI slop themes he began cataloging:
- ✅ "Kids in the third world doing impressive stuff," like a child in Africa building an elaborate statue out of trash
- ✅ Religious figures placed in absurd situations
- ✅ Military heroes in fabricated scenarios
- ✅ "Wholesome" content engineered to trigger an emotional reaction
He came to believe the creators behind this content had simply noticed that wholesome-looking fakes perform well, so they kept manufacturing more of it.
The Viral Effect
Théodore's account struck a nerve because millions of people were experiencing the same frustration.
They were seeing their feeds fill with obvious fakes. They were watching friends and family members share content that was clearly AI-generated. And they had no idea how to stop it or even talk about it.
The "Insane AI Slop" account gave them a name for the problem and a community to share their outrage.
His followers grew to over 133,000 within months, not because he was saying anything radical, but because he was saying what everyone else was already thinking.
You might also want to read: the darker side of AI influencer culture we've covered before
What Exactly Is "AI Slop"?
Merriam-Webster's Word of the Year
In 2025, Merriam-Webster officially named "slop" its Word of the Year, pointing to a flood of low-quality, mass-produced AI content, from absurd videos to fake news that looks pretty real, as the reason.
Think about that. A dictionary officially recognized that AI-generated garbage has become so prevalent it needed its own word.
The Evolution of Internet Content
To understand AI slop, you need to understand how we got here:
| Phase | Description |
|---|---|
| Phase 1 | User-generated content: friends, family, and real people sharing their lives |
| Phase 2 | Creator content: influencers and professional creators producing for audiences |
| Phase 3 | AI-generated content: algorithms producing content for other algorithms |
We're now in Phase 3. And it's not going well.
The Hallmarks of AI Slop
AI slop has distinct characteristics that make it recognizable once you know what to look for.
Visual tells:
- ✅ Unnatural proportions and impossible anatomy
- ✅ Glitchy backgrounds and melting details
- ✅ Strange lighting that doesn't match the scene
- ✅ Text that looks like gibberish upon close inspection
Content tells:
- ✅ Overly emotional or "wholesome" narratives
- ✅ Absurd scenarios that trigger outrage or sympathy
- ✅ Generic, stock-photo-style compositions
- ✅ Repetitive themes across multiple accounts
The "Shrimp Jesus" phenomenon, where Facebook was briefly flooded with AI-generated images of Jesus fused with crustaceans, became one of the most famous examples of AI slop going mainstream.
You might also want to read: our full breakdown of the telltale signs of AI slop
The Shocking Numbers: How Bad Is It Really?
The Statistics That Will Blow Your Mind
Let's look at the numbers. They tell a striking story:
- 40%+ of viral content on Facebook is estimated to be AI-generated
- 56% of consumers report seeing AI "slop" often in their feeds in 2026
- The share of friend-posted content on Facebook fell from 22% in 2023 to 17% in 2025
- Engagement rates on Facebook have collapsed to around 0.15% per post
- One AI content network published 355,296 posts in a single stretch, peaking at 3,919 posts in one day
- That same network's posts have drawn more than 95 million reactions, comments, and shares since 2025
The YouTube Problem
It's not just Facebook. In YouTube CEO Neal Mohan's 2026 letter to creators, he revealed that in December alone, more than one million YouTube channels used the platform's AI tools daily.
That's a million channels using AI to produce content in a single month.
The Real Cost
Meta's AI capital expenditures have climbed to a projected $125 to $145 billion for 2026, even as its user numbers dipped for the first time on record.
Think about that irony: Meta is spending billions on AI while users are leaving because their feeds are filled with AI garbage.
And yet, according to Meta's official first-quarter 2026 results, the company still posted $56.31 billion in revenue, up 33% year over year.
The platform is making more money than ever while delivering a worse product than ever.
You might also want to read: the data on how much AI slop is flooding YouTube Shorts too
The $300,000 Question: Who's Making Money From This?
The Global Slop Economy
Here's where it gets uncomfortable.
AI slop isn't just annoying. It's profitable. Extremely profitable.
An investigation by the Bureau of Investigative Journalism uncovered a cottage industry of young entrepreneurs in Pakistan and Sri Lanka using AI tools to produce deeply objectionable content, and making serious money doing it.
The Numbers Behind the Scam
- One Pakistani creator told investigators he makes about $1,500 a month from one Facebook page alone
- A Sri Lankan creator, Geeth Sooriyapura, claimed to have made $300,000 over the course of his Facebook career
- A Pakistani account posted screenshots claiming more than 1.2 billion views and roughly $21,600 in revenue over four months
How They Do It?
These creators aren't ideological. They don't care about politics or religion. They're entrepreneurs chasing the algorithm.
- They create pages that look like patriotic local news accounts
- They generate outrage-bait content using AI tools
- They drive engagement through emotional manipulation
- They profit from the ad revenue Meta places next to their content
Investigators traced several accounts pushing false narratives, including AI-generated Holocaust-victim imagery flooding Facebook, back to a small circle of "digital creators" in South Asia, part of a much larger economy of deception and emotional manipulation exploiting Meta's payment models.
The Dark Side of the Slop Economy
Some of this content isn't just low-quality. It's actively harmful.
Investigators found AI slop operations in South Asia producing Islamophobic content targeting British audiences.
One page with a Union Jack profile picture posted AI-generated videos claiming local cafes had stopped serving pork to avoid offending people, content that fuels far-right talking points.
Comments beneath these videos show users calling for mass deportations and mocking AI-generated footage of migrants drowning in the Channel.
This isn't just annoying content. It's real-world harm being monetized by platforms that won't stop it.
You might also want to read: the global AI slop economy we investigated in more depth
How AI Slop Actually Works?
The Playbook Revealed
AI slop isn't random. It follows a clear, repeatable playbook that anyone can learn to spot.
Step 1: Set up fake "local" pages.
Creators build pages that look like legitimate local news outlets. They use generic names, stock profile pictures, and a mix of real and fake content to build credibility.
Step 2: Generate emotional bait with AI.
Using tools like Midjourney, DALL-E, or Stable Diffusion, they crank out dozens of images per hour. The prompts always target high-emotion topics: tragedy, patriotism, religion, or manufactured poverty.
Step 3: Post relentlessly.
One network posted nearly 4,000 times in a single day. Volume beats quality. The algorithm rewards frequency, so they flood the zone.
Step 4: Cross-post to multiple pages.
The same image gets shared across 20+ different pages. If one goes viral, they all benefit. This creates the illusion of organic, grassroots sharing.
Step 5: Rinse and repeat.
Once a template works, say, a crying child holding a sign, they generate hundreds of variations with different backgrounds and text.
The Technology Behind the Slop
Most of these creators aren't tech wizards. They use off-the-shelf AI tools that are cheap and widely available:
- Image generation: Midjourney, DALL-E, Stable Diffusion
- Video generation: Runway, Pika, Sora
- Text generation: ChatGPT, Claude, Gemini
The barrier to entry is almost zero. Anyone with an internet connection and $20 can start generating slop within minutes.
Why Facebook Is the Perfect Host?
Facebook's algorithm is designed to maximize time on site and engagement. Emotional content, even fake emotional content, performs better than factual, neutral content.
- The algorithm doesn't care if a photo is real, only if people react to it
- Fact-checking is slow and reactive, while AI generation is fast and proactive
- Facebook's recommendation engine actively promotes content that triggers outrage or sympathy
As a result, Facebook has become the ideal platform for AI slop entrepreneurs.
You might also want to read: the specific playbook scammers use to drain bank accounts with AI
Meta's Response: Too Little, Too Late?
What Zuckerberg Says vs. What He Does?
Mark Zuckerberg has acknowledged the problem, sort of.
On a recent earnings call, he admitted AI-generated content would "crowd out" user-generated content and that Meta needed to think carefully about managing it.
But actions speak louder than words.
What Meta is actually doing:
- ✅ Expanding AI-generated profile suggestions and content recommendations
- ✅ Investing $125 to $145 billion in AI infrastructure this year
- ✅ Experimenting with AI-powered community notes, similar to X's system
- ✅ Adding watermarks to AI-generated images, though these are easy to remove
What Meta is not doing:
- ❌ Banning AI-generated pages outright
- ❌ Reducing ad revenue sharing for slop creators
- ❌ Prioritizing human-created content over AI content
- ❌ Implementing real-time AI detection at scale
The Community Notes Experiment
Facebook is currently testing a community notes feature, similar to what X uses.
The idea is that users can add context to misleading posts. In theory, this could help flag AI slop.
But there's a catch: community notes are reactive, not proactive. By the time a note appears, the slop has already gone viral and made money for its creator.
The Revenue Problem
Here's the uncomfortable truth that Meta rarely says out loud: AI slop is profitable for Meta too.
Every reaction, comment, and share generates ad impressions. Even if users are engaging with fake content, they're still seeing ads.
Meta pulled in $56.3 billion in revenue in Q1 2026 alone, and a significant portion of that comes from engagement-driven ad placements, the exact kind AI slop is built to drive.
So while Zuckerberg says he's worried about content quality, the balance sheet tells a different story.
You might also want to read: Meta's own "third phase" AI content strategy, straight from Zuckerberg
Why This Matters for You?
The Real-World Consequences
You might think it's just silly pictures. Who cares?
But the AI slop flood has serious, measurable consequences.
- Erosion of trust in all media. When your feed is 40% fake, you stop trusting anything you see. That cynicism spills over into legitimate news sources, making it harder to distinguish fact from fiction.
- Polarization and radicalization. Slop creators have learned that outrage sells. They deliberately generate content that inflames political and religious tensions, pushing users toward more extreme views over time.
- Wasted time and cognitive load. Your attention is valuable. Every second spent looking at AI-generated garbage is a second not spent with real content, real people, or real information.
- Ad revenue hijacking. Every dollar Meta pays to a slop creator is a dollar that could have gone to legitimate publishers, journalists, and creators. This actively harms the media ecosystem.
- The "dead internet" scenario. If AI-generated content keeps growing, we risk a tipping point where most online content is generated by machines, for machines, with humans as secondary.
What's at Stake?
The AI slop crisis isn't just a Facebook problem. It's a preview of what happens when AI generation outpaces AI detection and profit incentives override quality control.
If this goes unaddressed, expect the same pattern to spread further into:
- YouTube (already happening, with over a million AI-assisted channels)
- TikTok (AI-generated influencers and videos)
- Instagram (AI-generated models and lifestyles)
- LinkedIn (AI-generated thought leadership posts)
- News websites (AI-generated articles with little human oversight)
You might also want to read: why AI slop is especially dangerous for children online
How to Spot AI Slop?
A Practical Checklist for Everyday Users
You don't need to be a tech expert to spot AI slop. Use this checklist every time you see a questionable post:
- ✅ Look at the hands and fingers. AI still struggles with hands. Count the fingers and look for extra digits, missing fingers, or hands that blend into arms.
- ✅ Check the text. If there's any text in the image, signs, labels, banners, zoom in. AI-generated text is almost always garbled on close inspection.
- ✅ Examine the background. AI images often have backgrounds that melt into the foreground. Look for strange blobs, mismatched shadows, or objects that don't make sense.
- ✅ Test the proportions. Are the eyes too big? Are the teeth strange? Are the limbs impossibly long? These are classic AI tells.
- ✅ Ask if it makes logical sense. If a child is building a statue out of trash while wearing brand-new sneakers, something is off.
- ✅ Reverse image search. Run a screenshot through Google Images or TinEye. AI-generated images often return no matches, or identical copies on dozens of other pages.
- ✅ Check the source. Click on the page that posted it. Is it a generic name? Does it post 50 times a day? Those are red flags.
- ✅ Trust your gut. If something feels too perfect, too tragic, or too outrageous to be true, it probably is.
You might also want to read: how this same AI slop problem is driving users off Pinterest
The Future of Social Media
Where Do We Go From Here?
The AI slop rebellion has already changed the conversation. Théodore Cazals and his 133,000 followers have proven that people are paying attention, and they're not happy.
But awareness alone won't fix the problem. Here's what needs to happen.
For platforms:
- Implement real-time AI detection that flags and demotes slop
- Adjust algorithms to prioritize human-created, verified content
- Reduce ad revenue sharing for pages that post primarily AI-generated content
- Require clear, permanent labeling of all AI-generated media
For regulators:
- Introduce transparency requirements for AI-generated content
- Mandate that platforms report on the volume of AI slop in user feeds
- Fine platforms that knowingly profit from harmful AI-generated disinformation
For users:
- Stop engaging with AI slop; don't like, comment, or share it
- Report fake pages to the platform
- Follow journalists and investigators tracking this issue
- Demand better from the platforms you use
The Optimistic Scenario
It's not all doom and gloom. The fact that we're having this conversation, that a single student's account blew up and got global attention, shows that people still care about truth.
The same AI tools that generate slop can also be used to detect it. Researchers are already building AI models that spot AI-generated content with increasing accuracy.
The Pessimistic Scenario
If nothing changes, we're heading toward a "dead internet" scenario, where the vast majority of content is generated by AI, for AI, with humans serving mostly as passive consumers of algorithmic noise.
In that world, trust collapses, meaningful connection disappears, and social media becomes what its harshest critics always said it was: a distraction machine designed to extract attention and sell ads.
You might also want to read: how YouTube is now purging entire AI slop channels
Final Thoughts
Théodore Cazals started his "Insane AI Slop" account because he saw something absurd and couldn't look away. In doing so, he gave a name to a problem that was quietly eating social media from the inside.
AI slop isn't going away on its own. It's too profitable for creators and too convenient for platforms to moderate.
But now you know what it is, how to spot it, and why it matters.
The next time you see a crying child holding a sign, a bearded toddler in the rain, or a Shrimp Jesus floating through your feed, don't just scroll past. Pause. Look closer. Ask questions.
Because the only thing more powerful than an algorithm designed to manipulate you is a human being who refuses to be fooled.
Frequently Asked Questions (FAQ)
What is AI slop?
AI slop is low-quality digital content, usually images, video, or text, that is mass-produced using AI tools and posted for engagement or ad revenue rather than accuracy or value. Merriam-Webster named it the 2025 Word of the Year.
Who is Théodore Cazals?
Théodore Cazals is a French student who created the X account "Insane AI Slop" (@FacebookAIslop) to expose viral AI-generated fake content on Facebook. His account has grown to more than 133,000 followers.
How much money do AI slop creators make?
Reported earnings vary widely. Investigations found individual creators earning roughly $1,500 a month from a single Facebook page, with some claiming totals as high as $300,000 over their careers, largely through Meta's ad revenue sharing programs.
How can I tell if a Facebook post is AI-generated?
Look closely at hands, fingers, background details, and any text in the image, check if the scenario makes logical sense, and try a reverse image search. Generic-looking pages that post dozens of times a day are also a red flag.
Is Meta doing anything to stop AI slop?
Meta has said it is testing community notes and watermarking AI images, and is expanding AI-generated content recommendations. Critics point out that Meta has not banned AI slop pages, reduced revenue sharing for them, or rolled out real-time detection at scale.