Photo via androidguias
The feature is called Scam Alert. It analyzes messages from unknown contacts and looks for signs of fraudulent schemes. When the protection is triggered, the user will see a warning about a possible scam and can immediately block the sender and report them. If the system makes a mistake, the conversation can be marked as trusted.
The sender will not be notified that the protection was triggered—the warning is displayed only to the recipient.
Verification will take place directly on the smartphone
WhatsApp will verify suspicious messages directly on the smartphone. A specialized machine learning model analyzes the conversation locally, without transmitting its content to Meta’s servers or third-party cloud services.
The company has also provided the option to voluntarily share data. For example, if a user believes a warning was a false positive, the app may offer to send the developers the last five messages from the conversation to further train the model.
The app will also feature a local Scam Alert activity log. There, users will be able to see when the protection was activated, which messages were flagged as suspicious, and which version of the model was used for analysis.
For now, the new feature is in the early stages of testing. According to WABetaInfo, Scam Alert is available only to a limited number of users of the WhatsApp beta for Android and is disabled by default. Even installing the latest test build does not guarantee access—activation occurs on the server side.
In the future, Scam Alert may also become available on the iPhone, though WhatsApp has not yet announced a launch date for the feature.
