Meta has begun testing a new artificial intelligence feature on WhatsApp designed to protect users from fraudulent and scam messages, AzerNEWS reports.

Called Scam Alert, the feature identifies potentially suspicious messages and warns users about possible fraud attempts. It is currently being rolled out to a limited number of users as part of a beta test, and activation of the feature is optional.

Scam Alert uses an on-device machine learning model to analyze incoming messages. When the system determines that a message is likely to be part of a scam, it displays a warning within the conversation.

Only the recipient sees the warning, while the sender is not notified that the message has been flagged.

After receiving a warning, users can choose to block the sender, report the account to WhatsApp or continue the conversation. If a message is incorrectly identified as suspicious, users can mark the conversation as trusted. Scam Alert will then stop displaying warnings for that particular chat.

WhatsApp also gives users the option to share their five most recent messages with the company to help improve the feature’s accuracy. This data-sharing option is voluntary.