**WhatsApp** has begun rolling out an innovative new feature designed to protect its vast user base from the persistent threat of online scams. The optional 'Scam Alert' system utilizes a local machine learning model to detect and warn users when they might be targeted by fraudulent schemes. ### On-Device Protection for Privacy Currently in a limited beta rollout, the **Scam Alert** feature is being tested with researchers in **WhatsApp's** Bug Bounty community. A key design principle is privacy: "No message content leaves the device for classification or is auto-reported to **WhatsApp**, **Meta**, or anyone else," the company stated. This ensures that the feature complements **WhatsApp's** end-to-end encryption, offering a user-controlled, optional scam alert when the model identifies a likely scam. ### How Scam Alert Works The machine learning model is trained on a dataset of scam conversations previously reported by users. It analyzes incoming messages from non-contacts, looking for linguistic signals and conversational structures that match known scam patterns. If a potential scam is detected, users will see a prominent chat warning, prompting them to block, report, or continue the conversation. Users have the ability to mark a warning as incorrect, in which case the alert will be removed, and that specific chat will not be flagged again. For instances where a user trusts a chat, they can also opt to share the last five messages received with **WhatsApp** to help improve the feature's accuracy. **WhatsApp** emphasizes that all message data processed by the local model and the model itself never leave the device, and the feature can be disabled at any time. ### Broader Anti-Scam Initiatives This new initiative is part of **WhatsApp's** ongoing commitment to bolster user security against scammers. Previous updates have introduced various security features to preempt fraud attempts. In March, **Meta** announced that **WhatsApp** would warn users when behavioral signals indicate a potentially fraudulent device-linking request. This addresses a common tactic where scammers trick users into scanning malicious QR codes or sharing linking codes to hijack accounts. Two months prior, **WhatsApp** also rolled out "Strict Account Settings," a lockdown-style security feature aimed at protecting high-risk individuals such as journalists, public figures, and activists from sophisticated threats, including spyware attacks. This feature was developed in response to numerous incidents where zero-click exploits were used to infect iOS and Android devices via messaging apps like **WhatsApp**, often involving spyware like **NSO Group's Pegasus**. With over 3 billion users across more than 180 countries, **WhatsApp's** continuous efforts to enhance security are crucial in safeguarding its vast global community from evolving cyber threats.