Apple Announces CSAM Detection Plans
Published:
Apple announced a set of child-safety features in August 2021 that included a controversial plan to detect known child sexual abuse material in iCloud Photos before images were uploaded.
The proposed system used a perceptual hashing technology called NeuralHash. A device would compute identifiers designed to remain similar even when an image was modified, compare them against a blinded database of known illegal material, and create encrypted safety vouchers. Apple said human review would occur only after an account crossed a threshold of matches.
Security and privacy researchers objected that building client-side scanning infrastructure could create a precedent for detecting other material on personal devices. Apple delayed the feature and ultimately did not deploy the original CSAM scanning design. The debate became a major example of the tension between endpoint privacy and automated content detection.