Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteSome links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
There is no single bot or AI account behind Facebook’s bizarre viral images. A 404 Media investigation published in August 2024 traced examples to human-run pages and a small-scale creator economy: operators make cheap, attention-grabbing images, post them to Facebook, and try to earn money through Meta’s performance-based monetization programs. The images are generated by AI; the system that gives them reach and a possible payoff is human and commercial.
What “AI slop” means here
“AI slop” is a dismissive term for low-effort, formulaic material made primarily to capture attention rather than inform or express a considered idea. In Facebook feeds, it often means synthetic images built around an instantly legible emotional hook: devotion, pity, patriotism, shock, or outrage. The label describes a pattern, not every image made with AI. AI-generated art, satire, or clearly fictional entertainment is not automatically slop, spam, or misinformation.
The examples documented in reporting include “Shrimp Jesus” and other surreal religious scenes, distorted or starving-looking people, children or older people presented as objects of sympathy, and invented floods, fires, rescues, animals, trucks, or oversized objects. Some posts add blunt prompts for likes, comments, or shares. Such images can be visually arresting even when their anatomy or premise is plainly impossible. And a comment saying “this is fake” can still be an interaction that helps a post travel further.
Follow the production line
The 404 Media investigation described operators and aspiring creators associated with India, Vietnam, and the Philippines. That is a reported pattern in the people and examples the outlet investigated—not evidence that creators generally in those countries make spam, or that most Facebook AI content comes from them. It also showed that the work is not necessarily done by one person: image makers, page administrators, online instructors, guide sellers, and viewers can all play different roles.
#1 Best Overall
- Make an image. An operator enters a prompt into an image generator and tries themes likely to attract attention. The 2024 reporting identified Microsoft’s Image Creator among the tools used.
- Publish through a page. The operator posts to a Facebook page, sometimes one with a substantial audience. Futurism’s summary of the investigation cited examples of pages with more than 100,000 followers.
- Seek reach and reactions. The image is designed to invite a quick response—wonder, sympathy, disbelief, argument, or a request to share. Facebook recommendations can put page content in front of people who do not follow that page.
- Try to monetize performance. If the page and content meet the requirements of a relevant Meta program, the operator may seek a payout. Eligibility, enrollment, policy compliance, and Meta’s calculation all matter; posting a viral image does not guarantee payment.
This is a repeatable tactic, not proof of a centrally coordinated network. The investigation found human operators learning and applying a method; it did not establish that all such posts are automated, that every viral image belongs to the same operation, or that a particular creator made any image merely because it resembles a familiar template.
YouTube tutorials, Telegram guides, and a claimed rate
According to 404 Media, some operators learned the approach from YouTube influencers or bought guides through Telegram. The instructional material described in the reporting covered practical steps such as setting up a Facebook page, choosing image themes and prompts, uploading content, and pursuing Meta’s “Performance Bonus.” That turns casual experimentation with an image generator into a business recipe: produce cheaply, publish repeatedly, and hope engagement becomes income.
Futurism reported that a YouTube creator claimed earnings of roughly $3 to $10 per 1,000 likes. Treat that as an attributed creator claim, not a standard Meta price, an independently verified rate, or a promise that likes convert directly into cash. Rates and programs can vary, and the 2024 figure should not be assumed to describe what a creator can earn today. A later 404 Media follow-up, published April 15, 2025, described examples of creators earning hundreds of dollars from viral images and a broader market for disaster-themed material. Those were reported examples, not typical or guaranteed earnings.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Rank #2
- Enhanced Screen Recording - Capture screen & webcam together, export as separate clips, and adjust placement in your final project.
- Color Adjustment Controls - Automatically improve image color, contrast, and quality of your videos.
- Frame Interpolation - Transform grainy footage into smoother, more detailed scenes by seamlessly adding AI-generated frames. (feature available on Intel AI PCs only)
- AI Object Mask - Auto-detect & mask any object, even in complex scenes, to highlight elements and add stunning effects.
- Brand Kits - Manage assets, colors, and designs to keep your video content consistent and memorable.
What Meta’s rules say—and what the reporting found
Meta’s Content Monetization Terms require eligible content to comply with its terms and policies. They reserve the company’s ability to withhold payment for violations, fraud, or other legal issues. The terms describe payments as dependent on eligibility and Meta’s calculations; they do not say that every AI-generated image is banned. They also specify payment thresholds for certain payouts—$25 for U.S. residents and $100 outside the U.S.—but those figures should not be treated as rules for every Facebook bonus or monetization product.
Meta told 404 Media that many of the images at issue did not violate its policies, and said the program was working as intended where reach was not artificially boosted with bots. That is Meta’s position as reported by Futurism, not an independent finding about every post or page. It helps explain the gap: an image can be grotesque, misleading in implication, or aesthetically low-effort without necessarily breaking a clearly defined platform rule. Eligibility under a monetization program is not the same as a platform endorsing the image’s message or quality.
So “Meta pays people to make fake images” is too broad. The evidence supports a narrower conclusion: performance-based programs created a financial incentive that operators tried to exploit with inexpensive, engagement-oriented content. The reporting does not show that Meta commissioned the images or set out to reward deceptive material. It does show why a system that rewards performance can be attractive to people who can make many cheap posts and test which ones travel.
Why these images can spread
Three conditions reinforce one another. First, image generators make it inexpensive to produce unusual scenes. Second, Facebook’s recommendation feed can distribute page posts beyond their existing followers. Third, emotionally direct images can prompt reactions even when viewers are skeptical. A viewer may share an apparent rescue out of concern, argue over a religious image, or comment to correct a false impression. The same activity that expresses doubt can still register as engagement.
That does not prove that a recommendation system rewards every misleading image, or that every creator is consciously following a psychological playbook. It is a plausible mechanism for why reaction bait can gain distribution. A former Meta employee cited in the reporting argued that the ability to publish at scale creates opportunities to exploit platform weaknesses at scale; that is an attributed analysis, not a measured estimate of how much the practice contributes to Facebook’s feed.
There is also a moderation problem. Removing content based on defined policy violations is not identical to suppressing everything false-looking, strange, or unpleasant. A strict rule against synthetic imagery could sweep up harmless art, satire, or legitimate creative work; a permissive approach can leave room for material that is manipulative but difficult to classify. Labeling, automated detection, human review, and enforcement consistency each address different parts of that tension.
The image tool is not the distribution system
Microsoft’s Image Creator was one tool named in the 2024 reporting, not the source of the Facebook economy. Microsoft currently describes Bing Image Creator as a consumer image-generation service accessible with a Microsoft account. Its available models, creation modes, and usage limits can change. An accessible generator lowers the cost of making an image; Facebook pages, recommendations, audiences, and monetization create the separate distribution and incentive environment.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.AI labels do not settle whether a post is trustworthy
The available reporting does not support a claim that every synthetic image on Facebook is labeled—or that none is. The 2025 404 Media follow-up described creators being advised to disclose that posts were AI-generated, while noting uncertainty about whether an AI label alone was sufficient under Meta’s rules. Disclosure can tell a viewer something about how an image was made. It does not make an invented disaster real, make a humanitarian scene accurate, or determine whether the post is eligible for monetization.
Recommended Free Tools
It is useful to distinguish four cases that can look similar in a feed:
- Legitimate synthetic work: made for art, humor, or entertainment without pretending to document a real event.
- Engagement bait: formulaic or emotionally provocative material built to elicit reactions, which may be misleading or merely low-quality.
- Deceptive synthetic media: an invented scene presented as evidence of a real person, disaster, rescue, or event.
- Coordinated spam: repeated, bulk, or cross-page posting aimed at exploiting recommendation or monetization systems.
An ugly image alone cannot tell you which category applies. A page can have legitimate monetization status while a particular post is later removed or made ineligible; an image can be reposted so often that its original creator is hard to identify; and an AI label does not by itself establish whether the caption is true.
Meta’s wider turn toward AI content
In an October 2024 report, Futurism covered Mark Zuckerberg’s comments that Meta expected to add more AI-generated, AI-summarized, or AI-assembled material to Facebook and Instagram. That points to a broader strategy of algorithmically recommended and synthetic content; it is not proof that Meta endorsed fraudulent images or abandoned all moderation. It does, however, make the distinction between “AI content” and “AI slop” increasingly important: the technology may be built into platform experiences even as low-quality engagement bait remains a separate enforcement and incentive problem.
How to judge a suspicious post before sharing
- Check the claim, not just the pixels. If a caption says an image shows a current flood, rescue, or other real event, look for independent reporting or a reliable local source before sharing it.
- Inspect the page’s history. Repeated themes, nearly identical formats, and a high volume of unrelated emotional scenes may suggest an engagement-oriented publishing pattern. They do not prove who made an image.
- Look for context. Does the post name a location, date, or source that can be checked? A vague caption asking for likes or shares is not evidence that the scene is real.
- Treat labels as clues, not verdicts. An AI disclosure can explain synthetic origin, but it does not verify the accompanying story or establish that the post is harmless.
- Report deceptive content when appropriate. Use Facebook’s available reporting option if a post appears to misrepresent a real event or violate a platform rule. Reporting is not a guarantee of removal, and visual intuition alone is not a reliable authenticity test.
The strongest documented explanation is not that an AI has taken over Facebook. It is that people can use cheap generative tools, emotional hooks, page-based distribution, and performance-based monetization to make attention-seeking images worth trying. The images may be synthetic, but the incentive loop is recognizably human.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
