The Meme Factory: The Indonesians Making Money Off Facebook Outrage

The event is primarily a platform-economy story with AI as a production tool rather than the subject. Adira, interviewed for the piece, runs a 26,000-member pro-Pauli Hanson group, posts in Charlie Kirk and JD Vance fan groups, and shares AI-generated images such as a hijab-wearing girl holding a bacon-related sign and pictures of AI-generated pilots. She says she does not know who the US president is and that her only filter is which groups pay out. For an engineer, the AI angle is a cost-of-production fact: image generation collapses the effort needed to produce locally plausible content for any foreign niche.

The economic loop is Meta's creator program, which pays for engagement with videos, photos, and text posts. The company says it paid nearly $3 billion to creators in 2025. Researchers quoted in the piece argue this is what attracts workers who have no other capital than a phone and a few accounts, and that the term buzzer captures the way many small accounts add up to large ambient noise. The implication for builders is that any payout formula tied to engagement will be optimized by someone, and the optimization will move toward whichever topics currently monetize best, in this case, foreign political outrage.

Coordination happens off-platform. Adira describes Facebook and WhatsApp groups where Indonesian operators share which foreign groups are paying well and which content to post. Group discovery also runs through Facebook's own recommendations: she says suggested groups surface after she posts, and she picks new targets by chasing the ones that look monetizable. That is a concrete consequence for anyone designing a 'similar groups' surface: it is effectively a yield curve for engagement farming.

Where the system fails on its own terms is when operators slip, Adira posted Trump-focused Muslim content into an Australian Hanson group and acknowledges that 'people from overseas get suspicious.' Detection by audiences is one feedback loop; platform enforcement is another. Meta suspended seven of fourteen previously identified pro-Hanson groups for policy violations but did not address the payout structure that made the groups attractive in the first place. A researcher quoted in the piece argues the only durable fix is to end 'factory farming of meta-humans for profit,' which is a product-and-economics argument rather than a moderation one.

Only one feed carried this story, with a summary that offered no framing beyond 'Comments.' That thin pickup is itself a useful data point for an engineer thinking about signal routing: a piece about AI-generated political content can circulate as a generic social-media-moderation headline, which means downstream classifiers and editors need to look past surface tags to catch the AI-and-incentive structure. The grounding here is a single long investigation rather than corroboration, so any quantitative claim, the $3 billion figure, the seven-of-fourteen suspensions, rests on that one source.