Indonesian Political, Business & Finance News

Beware of Traps: New WhatsApp Feature to Detect Scam Messages

| Source: CNBC Translated from Indonesian | Technology
Beware of Traps: New WhatsApp Feature to Detect Scam Messages
Image: CNBC

WhatsApp is currently developing a new feature called ‘Scam Alert’, designed to detect suspicious messages from unknown numbers or contacts. If a message is flagged as a suspected scam, the sender will not be notified that their chat has received a warning.

This feature is now becoming available on a limited basis for WhatsApp beta users on Android. According to reports from WebetaInfo, Scam Alert is present in WhatsApp Android beta version 2.26.34.1, available via the Google Play Store.

Scam Alert will monitor incoming messages from unknown contacts. WhatsApp utilises an on-device artificial intelligence (AI) model to analyse messages and match them against recognised scam patterns.

If the system detects suspicious activity, WhatsApp will display a warning stating “This may be a scam” within the chat. This warning is only visible to the recipient, meaning the sender will remain unaware that their message has been flagged as a suspected scam.

WhatsApp provides users with several options following such an alert. Users can choose to block and report the contact, or mark the conversation as trusted to remove the warning and instruct Scam Alert not to flag that contact again.

The Scam Alert feature is optional and disabled by default. Beta testers can enable it through their WhatsApp account settings and can disable it again at any time.

WhatsApp has designed Scam Alert with conversation privacy in mind. According to Webeta $ ext{Info}$, the system uses a machine learning model that runs directly on the user’s device.

When a message from an unknown contact arrives, the model analyses the message content locally to determine if there are patterns resembling a scam. During this process, messages are not sent to WhatsApp, Meta, or any external servers for analysis.

Meta trains this model using patterns found in fraudulent conversations based on reports manually submitted by users to WhatsApp.

The company has also implemented several mechanisms to prevent the Scam Alert model from being used discriminatorily against specific users.

Furthermore, Meta is expanding its Bug Bounty programme for Scam Alert. This step allows external security researchers to test the system to ensure the model only detects scam patterns and is not used for any other form of analysis.

View JSON | Print