Scammers usually do not need much to start a conversation. An unfamiliar number, an urgent request or an offer that looks too good to be true can be enough to begin the process.
WhatsApp is now testing a feature designed to intervene at that first point of contact.
The messaging platform is testing Scam Alert, an on-device AI feature that can warn users when messages from people they do not know show patterns associated with potential scams.
The feature is currently being tested with a limited number of Android beta users on WhatsApp version 2.26.34.1. It is optional and switched off by default.
The interesting part is where the AI works
The biggest detail about Scam Alert is not necessarily that WhatsApp is using AI to identify scams. It is where that AI analysis happens.
The detection takes place directly on the user’s phone rather than requiring the contents of the conversation to be sent to WhatsApp’s servers.
That distinction matters because WhatsApp’s conversations are protected by end-to-end encryption. By performing the initial scam analysis on the device, the company can introduce an additional layer of protection without routinely sending the message content to its servers for analysis.
If the system detects suspicious patterns, it can display a warning inside the conversation. Users can then choose to block and report the sender, continue the conversation or mark the contact as trusted.
The sender is not notified that their message has been flagged.
WhatsApp is giving users a way to correct the AI
AI detection systems are not perfect, particularly when trying to distinguish between suspicious behaviour and legitimate conversations.
WhatsApp’s test therefore includes a way for users to override Scam Alert.
If a legitimate conversation is…
Source link
Read Full Article by Jessica Adiele at innovation-village.com
Source link
