How We’re Building Scam Alert on WhatsApp With End-to-End Encryption and Verifiability Guarantees
WhatsApp's on-device ML scam detector classifies messages without any content leaving the device
“The model is small enough to run on-device, simple enough to publish for independent review, and effective without server-side components.”
Meta's WhatsApp is rolling out an optional on-device machine learning feature called Scam Alert that classifies incoming messages from non-contacts as potential scams without any content leaving the device or being auto-reported. The system is designed to complement end-to-end encryption by keeping inference fully local, a meaningful demonstration of practical on-device ML for consumer-scale privacy-sensitive use cases. While not a foundational AI research announcement, it signals growing industry maturity around privacy-preserving on-device inference at scale.