WhatsApp’s new AI ‘Scam Alert’ can detect risky messages

WhatsApp is reportedly rolling out a new feature “Scam Alert” to tackle message scam in a new push for safety.

The new feature will be powered with artificial intelligence aimed at checking through messages from people who are not users’ contacts. By using “conversational structure and linguistic signals” the tool assesses and identifies whether a message is likely to be part of a scam.

After completing the assessment and identifying any sign of risk, Whatsapp will show a warning in the chat that will be invisible to the other person. After issuing the warning, it will still be the user's prerogative to decide whether to block or report the user sending it or carry on as normal.

Moreover, the system gathers data from previous conversations that WhatsApp flagged as possible scams. By analyzing the previous messages, it can flag the new one as normal or risky.

The users can switch on this tool manually. After switching on, the app runs on a machine learning model that is responsible for checking those messages. Addressing the privacy concerns, Meta stressed that this tool is designed to run on the device, meaning that none of the conversation is sent to WhatsApp and third parties.

According to the company, certain data signals are gathered to ensure the feature functions correctly and adapts to emerging scams. These signals are securely processed and transmitted to Meta in an entirely anonymized format.

For now, the feature will be rolled out in a limited beta version to early users.

WhatsApp parent company is planning to apply “stress-test” on tests to find bugs and vulnerabilities. After passing it, the feature will be available to all users globally.