A New York Times investigation published July 21, 2026 found that Meta’s AI-driven moderation systems, which now handle roughly 95% of first-pass enforcement decisions, have wrongly deleted legitimate Facebook and Instagram accounts — including business pages with large followings — after misclassifying ordinary content such as family photos and dance videos as child-exploitation or fraud violations. Appeals were routed back to the same automated systems that made the original error, and some accounts were restored only after journalists brought specific cases to Meta’s attention. Meta’s own Oversight Board said in June 2026 that the ban process lacks due process and disproportionately affects non-English speakers.