WhatsApp is now testing a new warning feature called Scam Alert: an AI model checks incoming messages from unknown senders for typical fraud patterns directly on the smartphone. If it detects an anomaly, a warning notice appears in the chat, visible only to the recipient. The feature is currently running only in a limited beta test.
How WhatsApp detects suspicious messages
Scam Alert relies on an AI model – a computer program that has learned to recognize typical patterns from many examples. Anyone who activates the feature downloads this model directly to their smartphone. It checks incoming messages from people who are not in the user’s contacts for phrasing and conversation patterns that often appear in fraud attempts. As Meta explains in a technical blog post, the message content never leaves the device and is not automatically reported to the company.
If the model detects an anomaly, WhatsApp displays a warning notice directly in the affected chat. This notice is visible only to the recipient; the sender does not see the warning and therefore cannot react to it specifically. An example from office life: if a supposed supplier writes from an unknown number and pushes for a quick transfer, Scam Alert can flag the message as suspicious before the case handler even responds. Similar scams involving cloned boss voices over WhatsApp have kept security authorities busy for months.
Users decide on the next steps themselves
When the warning appears, the decision rests entirely with the recipient: block the sender, report the message to WhatsApp, or continue the conversation anyway, for example if the contact is considered trustworthy after all. Anyone who permanently classifies a number as safe can mark the chat as trusted; future messages from that person will then no longer trigger a warning.
For transparency, WhatsApp keeps a log directly on the device showing when Scam Alert was triggered. According to the company, only anonymous metrics such as the number of warnings are sent to Meta, never conversation content. For office users, this means the feature does not replace personal caution but adds an automatic hint before reacting to an unusual payment request or link. It remains important to verify urgent money requests independently through a known phone number. Apple’s security framework Trust Insights works similarly, searching for fraud patterns entirely on the device since iOS 27 without transmitting message content.
Only a handful of beta testers have access so far
Scam Alert is currently in a limited beta test for Android and iOS, as the developer portal WABetaInfo found in recent test versions of the app. Anyone who does not see the feature simply does not have access yet – Meta itself decides which test groups get it first. There is no list of participating countries yet, nor a date for a wider rollout; the company has not commented on a possible launch in Germany or the EU either.
There are no costs involved: Scam Alert is part of the already free WhatsApp app, and no additional subscription is required. An existing WhatsApp account is enough; no new sign-up is needed. Meta announced the feature on August 12, 2026, and said it would keep testing it together with its own bug bounty community before a wider release follows. Until general availability, affected users can only stay alert and check or report suspicious messages manually as usual.
What will matter is how quickly Meta expands the beta beyond the current test group, and whether Scam Alert gets German-language interfaces along the way. Until then, protection remains a matter of luck: anyone receiving a suspicious WhatsApp message today from a customer, supplier, or supposed superior cannot rely on an automatic warning and should verify unusual payment requests through a known channel regardless. Those using WhatsApp for business customer contacts should also follow internal rules on handling customer and company data, since chat histories can end up outside their own IT infrastructure depending on settings.
