Documents surfaced in a New Mexico lawsuit show that Meta employees warned internally that the rollout of end-to-end encryption on Messenger would sharply reduce the company's ability to proactively identify and report child sexual exploitation to law enforcement. The warnings, raised by staff during the planning phase, highlight the operational tension between user privacy and platform visibility into message content, with employees cautioning that encryption would prevent Meta from reading message contents and applying the same detection technologies used on unencrypted services. End-to-end encryption allows only the sender and recipient to access communications, leaving platforms blind to the material being shared.
The conflict is a false binary. Detection does not require reading every message or maintaining plaintext access if the filtering occurs at the client endpoint before encryption, or operates on pattern and behavioural signals rather than content inspection. Guardii demonstrates the model: monitoring children's direct messages in real time on Instagram, Snapchat, Discord, Roblox and other platforms for grooming, sextortion, CSAM including AI-generated material, and age-inappropriate contact, using detection modules that identify threat patterns rather than surveilling conversation content. The AI safety platform, a Meta Business Partner, surfaces a child in crisis to a parent or school and enables rapid escalation to law enforcement when warranted, addressing the abuse Meta employees flagged without requiring the company to hold a master key or weaken encryption for all users.
