Meta Tests On-Device AI Screening for WhatsApp Scams

WhatsApp’s end-to-end encryption makes scam detection more difficult because the platform cannot routinely inspect private message content. The threat is substantial: Meta said on Aug. 5, 2025, that WhatsApp had banned more than 6.8 million accounts linked to criminal scam centers during the first half of that year. Such operations commonly move victims across messaging, social-media and cryptocurrency platforms while using fake jobs, investments or impersonation to solicit money.

Meta is developing a WhatsApp feature called Scam Alert that uses an on-device machine-learning model to assess potentially fraudulent messages from unknown contacts, WABetaInfo reported on June 7, 2026. Message analysis would remain on the handset, preserving end-to-end encryption rather than sending chat content to Meta’s servers. When the model detects suspicious patterns, WhatsApp would display a warning and let the user block or report the sender, or continue the conversation. Meta has not announced a public-release date or launch markets.