WhatsApp Tests New Smart Shield to Combat Fraudulent Messages

Meta is moving to bolster WhatsApp users’ protection through a new optional feature called Scam Alert. According to The Verge, the feature relies on machine learning running on the user’s device to detect messages that may indicate fraud attempts, while leaving the final decision in the account holder’s hands.

The feature is currently being rolled out in a limited beta. When a potentially suspicious message is detected, an alert appears within the chat that is visible only to the recipient, who can then choose to continue the conversation, block the account, or report it. Users can also mark the conversation as safe if they believe the warning was inaccurate.

Meta states that message analysis occurs on-device, and message content does not leave the user’s phone for classification purposes. No automatic report is sent to the company. Users may optionally share their last five received messages to help improve the system’s accuracy.

The move comes amid rising reported losses linked to fraud on social media platforms, which exceeded $2.1 billion in 2025, according to data from the U.S. Federal Trade Commission.