WhatsApp has begun rolling out Scam Alert, an optional limited-beta feature designed to warn users when a message looks like a likely scam, according to BleepingComputer and Digital Trends. The feature is being tested as an early warning layer inside chats rather than as a server-side content review system. The key design choice is where classification happens. BleepingComputer reports that Scam Alert runs a local machine learning model on the user’s device, while Digital Trends describes the same approach as on-device artificial intelligence. In both accounts, the model evaluates incoming messages from people outside the user’s contacts for signals associated with known scam patterns, including language cues and conversational structure. WhatsApp says the system is meant to preserve the privacy model of end-to-end encrypted messaging. According to both outlets, message content does not leave the device for classification and is not automatically reported to WhatsApp, Meta, or any other party. BleepingComputer quotes WhatsApp saying the feature is user-controlled and optional, and that it is designed to complement end-to-end encryption rather than bypass it. When the model identifies a likely scam, the user sees a warning inside the chat. From there, both reports say the user can block the sender, report the conversation, or continue. The sender does not receive the warning, according to Digital Trends. The feature also includes a correction path for false positives. If a user decides a warning is wrong, they can mark the chat as trusted, after which Scam Alert will stop flagging that chat, according to both outlets. Users who mark a chat as trusted can also choose to share the last five received messages with WhatsApp to help improve the feature’s accuracy; the reports describe that sharing as optional. The rollout fits into a broader set of WhatsApp security features aimed at fraud and account compromise. BleepingComputer notes that Meta previously announced warnings for suspicious device-linking requests, a tactic used in account hijacking, and earlier introduced Strict Account Settings for higher-risk users such as journalists and public figures. Digital Trends also frames Scam Alert as an extension of Meta’s earlier work around suspicious device-linking requests. The scale is material because WhatsApp is used by more than 3 billion people in over 180 countries, according to BleepingComputer. Digital Trends separately cites Federal Trade Commission figures saying victims reported $425 million in losses from scams on WhatsApp and $2.1 billion across social media scams in 2025. Those loss figures are single-source in this cluster, but they explain why a privacy-preserving detection layer inside a messaging app is operationally important. Who benefits: WhatsApp users who receive scam messages from unknown contacts stand to benefit most, especially if warnings appear before a conversation escalates. Meta also benefits if it can reduce scam abuse without weakening its privacy claims. Who's exposed: Scammers using WhatsApp outreach to initiate fraud face more friction if the model reliably detects their message patterns. The main product risk is false positives, which WhatsApp is addressing through the trusted-chat option and optional feedback sharing.