Meta adds AI-based scam alerts to WhatsApp

Meta is adding an on-device machine learning-based fraud alert feature to WhatsApp that will warn users about suspicious messages. The feature, called “Fraud Alerts,” is currently in limited beta testing. When the AI model identifies an incoming message as a likely scam attempt, the user sees a warning directly in the chat, which is not visible to the other party. The user can then decide whether to block the sender, report them, or continue the conversation. If the user believes the warning is incorrect, they can mark the chat as trusted, and that conversation will no longer be checked. In that case, they can also send the last five messages to WhatsApp administration to improve the alert system. In 2025, American citizens lost over $2.1 billion to scammers on social media, with $425 million of that attributed to WhatsApp. The messenger has over 3 billion users. When the fraud detection feature is enabled, “message content does not leave the user's device for classification,” and nothing is automatically sent to WhatsApp, Meta, or anyone else. The feature can be disabled at any time.