Prime Video uses Amazon's computer vision service, Rekognition, to automatically scan video assets for explicit, violent, or otherwise restricted content — such as adult material, drug use, hate symbols, and graphic violence — before that content is processed or published on the platform.

Amazon Rekognition Content Moderation analyzes images and video frames, returning hierarchical category labels with confidence scores for each detected category. Within Prime Video's Media2Cloud asset management pipeline, moderation labels are applied as part of the automated metadata enrichment step. The system is designed so that AI handles the large majority of content — flagging up to 95% of unsafe material automatically — while human moderators review only the edge cases that require judgment.