Meta is testing an optional AI-powered scam alert feature on WhatsApp that warns users when conversations may involve fraud.
The limited test, reported by The Verge, displays a warning banner when AI detects that an active conversation could be part of a scam.
Users can choose to block the contact, report the account or continue chatting.
If the warning is incorrect, users can mark the conversation as trusted to dismiss the alert for that specific thread.
WhatsApp also allows users to share the last five messages of the conversation to help improve the tool’s accuracy.
The rollout follows Meta’s launch earlier this year of an automated scam detection feature designed to flag suspicious linked-device requests on WhatsApp.
