Creator's sue Facebook over broken moderation algorithms (www.engadget.com)

Top Facebook creators are suing Meta in numbers of small-claims cases after widespread demonetization, missed payments and opaque moderation notices they blame on broken automated systems. Mel Bouzad, a prolific page operator who says his network once earned $10k–$20k/month (with a peak of $68k in bonuses in one month), has filed dozens of suits — 23 related to other creators’ pages and multiple personal claims — to force reviews after pages were flagged for vague reasons like “limited originality” or being in an “ineligible country.” Other creators report account misclassification (a U.S. bank mistakenly tied to Malta), locked tax/payout fields, and months of unresolved unpaid earnings (examples include $11k and $16k stuck), despite paying for Meta Verified support and submitting IDs. For the AI/ML community this case highlights real-world harms from over-reliance on automated moderation and operational pipelines. Key technical issues: high false-positive rates for content-origin and originality detectors, brittle heuristics that infer geographic or legal status from noisy signals, and poor integration between moderation models and human escalation/appeals. Implications include economic risk for creators, increased legal pressure on platforms, and the need for better model transparency, provenance/watermarking for AI-generated media, human-in-the-loop review for high-stakes decisions, and robust support workflows that surface and correct systematic model errors. Meta disputes the claims, but the disputes underscore that model deployment and support engineering are as consequential as model accuracy itself.