Meta’s rapid shift to AI moderation is raising eyebrows among employees, who claim the models still remove harmless content and lack adequate oversight. Despite the company’s claims of improved accuracy, tests show language models make 13% fewer errors than humans but catch 10% more actual violations. This transition is already leading to layoffs, and employees are worried about the potential consequences. Why it matters: This rapid rollout highlights the need for more transparency and accountability in AI decision-making, particularly when it comes to content moderation.