
Illustration of WhatsApp’s tested Scam Alert appearing inside a chat from an unknown sender.
Meta is testing an optional WhatsApp feature called Scam Alert that warns users when an active chat may be fraudulent, then offers three immediate choices: block, report, or continue. The feature is still limited to a small beta, not a general rollout. WhatsApp says the system runs on the phone and is aimed at chats from senders who are not saved in the user’s contacts, making it a narrower screen for unknown inbound messages rather than a blanket fraud shield across every conversation.
Scam Alert in WhatsApp chats
Scam Alert is designed to appear inside an active conversation when WhatsApp determines that a chat may be suspicious. The warning is shown to the recipient, not the sender, and it is built around a simple decision flow: stop the chat by blocking the account, escalate it by reporting it, or keep going by choosing continue. That makes the feature a direct in-thread checkpoint, not a background account notice or a general safety reminder elsewhere in the app.
For users who rely on WhatsApp for sales, support, or customer follow-up, the practical value is straightforward: the warning is meant to interrupt action before money, credentials, or internal information are shared with an unknown sender. But it also has a clear limit. Scam Alert is a prompt to scrutinize a live conversation, not proof that every risky message will be caught and not protection for all existing chats.
On-device screening and the unsaved-contact limit
WhatsApp says Scam Alert works by downloading a small AI model to the device and checking messages locally. For the detection step itself, message content does not leave the phone. That is a meaningful detail for a product inside an end-to-end encrypted messaging app, because the screening is being positioned as something that happens on the handset rather than in the cloud.
The main boundary is just as important as the technical design: the system only checks chats from senders who are not saved in the user’s contacts. Messages from existing contacts sit outside this scam-detection flow. For operators and teams that use WhatsApp heavily, that means Scam Alert does not replace internal verification steps for payment requests, urgent account-change instructions, or other sensitive asks that could arrive through a saved contact. Unknown senders get the extra screen; known contacts do not.
Beta status, trust controls, and what remains unclear
The feature remains in a limited beta or test rollout, and Meta has not confirmed a broader launch date or a country-by-country availability list. It is also unclear which platforms or user groups are included in the current test. For now, the safest practical reading for most users is that Scam Alert is something to watch, not something they should assume is already live across their organization’s devices.
WhatsApp does give users a way to correct the system when a warning is wrong. A flagged chat can be marked as trusted, which lets the recipient dismiss a false positive instead of treating every alert as final. Users can also choose to share the last five messages from a flagged conversation to help improve the feature’s accuracy. That share is optional, not automatic.
Scam Alert follows an earlier WhatsApp anti-scam step introduced this year for suspicious linked-device requests. The difference is where the check happens and what it watches. The earlier protection focused on account-linked activity, while Scam Alert is aimed at active chats with unknown senders. Together they show Meta adding more scam checks around WhatsApp use, but this new test is still narrowly scoped to stranger messages and still not broadly available.
Disclaimer: This article was created with the assistance of AI. Images are for illustrative purposes only.
About the author

Samarth Agrawal is an AI and technology professional who writes about WhatsApp, automation, and emerging AI trends. He focuses on simplifying complex tech updates into practical insights for businesses, creators, and everyday users
