WhatsApp is testing an AI-powered Scam Alert tool to detect suspicious messages: Here is how it will work

HIGHLIGHTS

The feature analyses messages from unknown senders without automatically sending private chats to Meta.

Users can block, report, dismiss or trust a sender when WhatsApp flags a potentially fraudulent conversation.

Scam Alert is currently being tested with a small group of users, with no public rollout date confirmed.

WhatsApp is testing an AI-powered Scam Alert tool to detect suspicious messages: Here is how it will work

Meta is testing a new Scam Alert feature on WhatsApp that will use an on-device AI model to identify messages that can be linked to scams. The optional feature is currently available to a limited number of beta testers and is designed to detect suspicious patterns without sending users’ private conversations to WhatsApp or Meta. This comes as the number of scams on messaging and social media platforms continues to cause financial losses.

Digit.in Survey
✅ Thank you for completing the survey!

How WhatsApp’s Scam Alert feature will work

Users will have to manually enable the Scam Alert from WhatsApp settings. Once they enable it, the required machine learning model will be downloaded to the device and analyse incoming messages from unknown senders.

This system will look for patterns commonly associated with scams, including requests for personal information, attempts to create urgency and messages designed to build trust before asking for money. This can include romance baiting scams, where fraudsters establish fake emotional relationships before seeking financial assistance.

Also read: Google introduces Gemini 3.7 Flash AI model, claims it outperforms Claude Sonnet 5 

If WhatsApp detects a potentially suspicious conversation, it will display a warning banner to the recipient. Users can then block or report the sender, dismiss the warning, or mark the contact as trusted. Choosing the trusted option removes the warning and stops further alerts for that conversation.

Meta says private messages remain private

Meta has said that the Scam Alert model processes messages on the user’s device rather than sending them to its servers for analysis. This means the company does not automatically receive the conversations being checked by the system.

The instant messaging platform is also testing an optional way for users to help improve the AI model. Users can choose to share the last five messages from trusted senders with WhatsApp. This data is not sent automatically and requires the user to actively opt in. It must be noted that the Scam Alert feature will remain in limited beta testing and Meta has not revealed when it will be available to all WhatsApp users.

Ashish Singh

Ashish Singh

Ashish Singh is the Chief Copy Editor at Digit. He's been wrangling tech jargon since 2020 (Times Internet, Jagran English '22). When not policing commas, he's likely fueling his gadget habit with coffee, strategising his next virtual race, or plotting a road trip to test the latest in-car tech. He speaks fluent Geek. View Full Profile