Behind the YouTube influencers, WhatsApp and Telegram groups, and services like FewFeed fueling Facebook's problem of creators profiting from viral “AI slop”
Gyan Abhishek is standing in front of a giant touch screen, like Jim Cramer on Mad Money or an ESPN talking head analyzing a football play. Threads: @jasonkoebler and @jenvalentino_nyt X: @jason_koebler , @jason_koebler , @jason_koebler , @jason_koebler , and @jason_koebler Threads: Jason Koebler / @jasonkoebler : After months of digging and reporting, I have learned where Facebook's bizarre AI spam (like “Shrimp Jesus") comes from, who is making it, how it works, and how it is monetized. Turns out Meta is directly paying people to spam FB with this stuff https://www.404media.co/... Jennifer Valentino-DeVries / @jenvalentino_nyt : Amazing work getting to the bottom of these creepy, annoying AI spam posts on Facebook. Meta has completely brought this on itself and is actually incentivizing these posts directly, apparently paying $100 for 1,000 likes. Way to junk up the internet, I guess. X: Jason Koebler / @jason_koebler : I also spoke to an Indian teenager who has been making vlogs about his efforts to make money spamming FB. He has a journal of 100 usernames/PWs he's written down, and a checklist for each task. He ultimately hasn't made any money yet [image] Jason Koebler / @jason_koebler : One person doing this made $431 from a single image of a train made out of leaves. This fact made him a celebrity on the “how to spam Facebook for money” podcast circuit. Used the money to buy his disabled brother an autorickshaw [image] Jason Koebler / @jason_koebler : While digging into this, I found a tool called FewFeed, made by a single Filipino coder, that costs $5 for two months and automates FB spamming and has features to, for example, seize groups that have admin openings. Can control dozens of accounts and pages at once [image] Jason Koebler / @jason_koebler : The images are weird because: - They are copypasta prompts from Telegram groups OR - They are speech-to-text in Hindi, run through Google translate. Translations are bad (or generated directly from Hindi) AND - They are a best guess of what will play with American audiences [image] Jason Koebler / @jason_koebler : Here's what I learned: - Facebook is paying for virality directly via its “Creator Program Bonus” - Influencers in India, Philippines, Vietnam have picked up on this and are making YouTube courses / Telegram classes - Images made with Bing AI Image Creator
Context & Ripple Effects
Facebook’s exposure to low-quality, recycled material predates generative AI: a 2018 internal memo on plagiarized and recycled Pages found that such Pages captured a substantial share of Page traffic. The reported network shows how image generation, translation, automation and creator payouts can now make that same engagement-gaming playbook cheaper to operate at scale.
The incentives also sit uneasily beside Meta’s earlier shift of resources toward the creator economy. Subsequent scrutiny of AI content on its apps, including Meta’s removal of experimental AI profiles after backlash, underscores that distribution and monetization policies can amplify content-quality problems before enforcement catches up.
First-order effects
- Creators using FewFeed, coordinated chat groups and AI-image tools can turn high-volume Facebook posting into a monetizable workflow, while Meta’s Creator Program Bonus rewards the engagement those posts attract.
- Facebook users and legitimate creators face a feed with more low-quality synthetic posts competing for attention; Meta must identify coordinated, multi-account behavior rather than judge isolated posts alone.
Second-order effects
- Meta’s payout and ranking systems become an immediate target for automation vendors and training communities, raising the cost and complexity of moderation, payment review and account enforcement.
- Creators producing original work may face tougher competition for reach and bonus eligibility from low-cost synthetic supply, while image-generation and translation tools become inputs to spam operations rather than only creative production.
Third-order effects
- If engagement-linked payouts continue to reward easily generated material, social platforms may need to treat monetization design, account provenance and coordinated-behavior detection as one integrity system rather than separate creator and safety functions.
- The pattern points to a broader synthetic-supply problem: generative tools lower production costs faster than platforms can distinguish useful creator output from industrialized engagement farming.
The trend: Generative AI is turning longstanding social-media engagement gaming into a more automated, cross-border content-production business tied directly to platform monetization.