WhatsApp is reportedly testing a new AI-powered security feature designed to help users identify potential scams while chatting. The feature could automatically analyse suspicious conversation patterns and display a warning when it detects signs commonly associated with fraudulent messages.

The proposed Scam Alert feature reportedly relies on on-device machine learning, meaning the analysis would take place directly on the user's smartphone rather than routinely sending private message content elsewhere for verification.

The feature is currently said to be available only to a limited number of beta testers and would remain optional for users.

WhatsApp Could Warn You About Suspicious Conversations

Under the new system, WhatsApp would analyse messages on the device to determine whether a conversation shows characteristics associated with scams.

If suspicious activity is detected, the app may display an alert directly inside the conversation. The warning would be visible only to the recipient, meaning the person on the other side of the chat would not be notified that their conversation has been flagged.

After receiving an alert, users could decide what to do next. Options may include blocking the contact, reporting the conversation or continuing the chat if they believe it is genuine.

False Warning? Users May Be Able to Mark Chats as Trusted

Automated scam detection isn't always perfect, and legitimate conversations could occasionally be flagged incorrectly.

To address this, WhatsApp is reportedly developing an option that allows users to mark a conversation as trusted. This could remove the warning and prevent the same conversation from repeatedly being highlighted as suspicious.

Such an option could be particularly useful when talking to new contacts, businesses or people whose numbers aren't already saved.

Sharing Recent Messages Would Be Optional

The feature may also allow users to voluntarily share their five most recently received messages with WhatsApp.

According to the reported details, this would be optional rather than automatic. The information could potentially help improve scam detection or provide additional context when users choose to report suspicious activity.

Users who do not want to share these messages would not be required to do so.

Privacy Remains an Important Part of the Feature

Privacy appears to be a major focus of the proposed scam detection system.

When the feature is enabled, the initial analysis of messages is expected to happen on the device. Simply receiving a scam warning would also not mean that the conversation has automatically been reported to WhatsApp, Meta or another party.

Users would remain in control of whether they want to take further action.

The scam detection feature is also expected to be optional, allowing users to disable it if they don't want conversations to be analysed for potential fraud.

How the Feature Could Help WhatsApp Users

Online scammers frequently use messaging platforms to impersonate banks, delivery companies, government agencies, employers or even friends and relatives. Fraudulent conversations may attempt to convince victims to share OTPs, passwords, banking details or send money.

An on-device warning system could provide users with an additional layer of protection by prompting them to reconsider a suspicious conversation before taking action.

However, automated alerts should not replace basic precautions. Users should still avoid sharing OTPs, PINs and passwords, opening suspicious links or transferring money solely on the basis of an unexpected WhatsApp message.

For now, the feature remains under limited beta testing. Its final functionality could change before a wider rollout, and WhatsApp has yet to make it available to all users.