WhatsApp tests scam detection as malware campaigns target users

WhatsApp is testing a new feature called ‘Scam Alert’ designed to identify fraudulent messages using on-device machine learning. The optional tool analyzes linguistic signals and conversation structures to detect potential scams from unknown contacts. To protect privacy, the analysis occurs locally on the user's device, and message content is not sent to Meta or used for classification unless a user explicitly consents to share messages to improve system accuracy.

In a separate cybersecurity development, Kaspersky’s Global Research and Analysis Team (GReAT) has identified a massive malware campaign targeting WhatsApp Desktop and WhatsApp Web users. Attackers are using compromised accounts to send malicious VBScript files to existing contacts. These files are disguised as routine business documents, such as invoices or bank statements, and are localized in multiple languages including English, Portuguese, French, German, and Malay. Once installed, the malware allows remote system access by mimicking legitimate IT support functionalities.