Etsy uses machine learning to automatically scan listings for policy violations — including prohibited items, hate speech, and counterfeit goods — before a human reviewer ever sees them. As of 2021, roughly 95% of flagged listings were identified by automated systems rather than human reports.

Etsy's Trust & Safety team uses supervised machine learning models that analyze both listing text (titles, descriptions, tags) and images to assign violation risk scores and surface high-priority content for human review. The text analysis component uses transformer-based language models (BERT and ALBERT), while image analysis uses convolutional neural networks (CNNs). In 2024, Etsy began extending enforcement to additional platform surfaces using large language model-based detection. Etsy's 2024 Transparency data showed a 70% improvement in automated detection precision, 3.5 million spam accounts banned (a 9x year-over-year increase), and 25% fewer total listing removals for policy violations compared to 2023 — reflecting improved accuracy rather than reduced enforcement.