WhatsApp Tests Scam Alert Feature to Detect Fraudulent Messages

WhatsApp is rolling out a new security feature designed to combat the growing threat of online scams. Dubbed "Scam Alert," the tool aims to empower users by helping them recognize potentially fraudulent messages directly within the app.

The feature operates using machine learning models that run locally on a user's device. This approach ensures that message content remains private, complementing WhatsApp's existing end-to-end encryption. According to Meta, WhatsApp's parent company, the models are designed to identify patterns associated with known scam tactics without needing to send message data to external servers.

Scam Alert is an optional feature that users must activate. Once enabled, a compact machine learning model is downloaded to the device. This model then analyzes incoming messages from individuals not saved in a user's contacts. It looks for communication patterns that align with previously identified scam methodologies, classifying messages probabilistically rather than relying on specific keywords.

Meta stated that the models are trained on scam patterns previously reported by WhatsApp users. The classification process considers conversational structure and linguistic signals to detect suspicious activity. This method allows the system to identify potential scams based on how a message is constructed and its context, rather than just isolated phrases.