Meta is testing an optional WhatsApp feature called Scam Alert, which uses machine learning on a user’s device to identify messages that may be fraudulent. It matters because WhatsApp’s huge user base makes it a major channel for scams, while the feature is designed to offer warnings without automatically sharing or reporting private messages.

The limited beta displays a warning visible only to the recipient, who can block, report or continue the conversation. Users can mark a wrongly flagged chat as trusted so it will not be screened again, and may optionally share the last five received messages to improve accuracy; Meta says message content otherwise remains on the device. The FTC recorded $425 million in reported WhatsApp scam losses in 2025, from more than $2.1 billion lost through social-media scams overall.