Bluesky uses an automated AI system supplied by a third-party vendor to scan every image and video frame uploaded to the platform and apply content labels such as adult, suggestive, nudity, graphic-media, and self-harm.

Every image and every frame of every video posted to Bluesky is sent to an external provider, which returns classification verdicts matched to Bluesky's label categories. These labels are then applied automatically. According to Bluesky's 2024 Moderation Report, this automation produced 5.5 million labels in 2024; by 2025 that figure had grown to 16.49 million labels, with 95% of all labels being automated. Bluesky states that neither it nor the vendor retains images or uses them to train generative AI systems.