Many people believe that when conversation is end-to-end encrypted, the problem is unsolvable. This incorrect belief is used by unscrupulous lobbyists and politicians who attempt to mandate various types of content scanning under the guise of preventing CSAM distribution.

We wrote before about how such measures not only would fail to solve the problem, but would make it worse. If our private photos become available to service providers, they will eventually become available to criminals too, and will be used to abuse and exploit the users and their children.

Apple attempted to solve this problem with NeuralHash but “failed” in the sense collisions could be easily generated. Has there been any subsequent work on privacy-affirming client-side scanning?