Spotify uses machine learning models to automatically scan music, podcasts, and audiobooks for policy violations — including hate speech, graphic violence, and child safety content — before and after content goes live.

System combines ML models (text, audio, image, video analysis), heuristic keyword/hash matching, and human review teams. Uses Thorn's Safer technology for CSAM detection. Per Spotify's EU DSA Transparency Report (February 2025), the automated system's error rate is 0.28%. Enforcement includes removal, restricted discoverability, restricted monetization, advisory labels, and country-level geo-restrictions.