Scammers are becoming more inventive by the day, attempting to slip through any crack to nab your personal data and, in most cases, your money. WhatsApp is no stranger to scummy scammers, even if you’ve been lucky enough not to come face-to-face with one. Today, Meta announced an “early look” at Scam Alert, a new detection method that runs on — you guessed it — artificial intelligence. It’s designed to fend off bad actors.
WhatsApp’s got your best interests at heart
If you’re confident in your ability to sniff out a scam before you hand over personal details or banking details, you can go ahead and leave Scam Alert turned off. But the elderly and gullible might want to take advantage of the on-device machine learning model, which is how WhatsApp claims to root out an incoming scammer.
“Before we make this feature available to all WhatsApp users, we are publishing this early technical overview alongside the feature’s limited rollout in Beta, and will continue working with our Bug Bounty community to stress-test this system,” the company wrote.
Read More: WhatsApp is finally giving iPhone users an Android-sized search upgrade
To make Scam Alert compliant with Meta’s end-to-end encryption, WhatsApp makes it clear that the machine learning model used here is entirely on-device and not on Meta’s servers. Turn the feature on, and it’ll automatically download the model. It’ll scan incoming messages from anyone not saved as a contact. From there, it’ll use its smarts to detect “conversational structure and linguistic signals” indicating a scammer.
If the AI model finds any particular message unsavoury, it’ll alert the receiver inside the chat with a warning — invisible to the sender. It’ll offer options to block, report, or even continue the conversation. If the user feels safe enough to proceed, they can flag the chat as trusted. When that’s chosen, they won’t be bothered by the warning again.
