Meta has launched a new tool within WhatsApp to curb the growing number of scams circulating daily on the app. Called Scam Alert, this feature promises to be a game-changer in combating digital fraud, as it uses an artificial intelligence model that works directly on the user's phone. The company has confirmed that the tool is currently in a limited beta phase, being tested with a small group of users before its general release.
The arrival of this feature is no coincidence. Scams via instant messaging have skyrocketed in recent times, and WhatsApp, with over 3.000 billion users, has become one of the main battlegrounds for cybercriminals. From identity theft and fake job offers to the well-known investment scams and romance scams, the methods are becoming increasingly sophisticated. With Scam Alert, the platform aims to take a step forward and offer an additional layer of protection without compromising the privacy of conversations.
How does the new scam alert work?
The way Scam Alert works is, to say the least, unusual. The tool is optional, and once enabled, it downloads a machine learning model onto the device. From that moment on, the system analyzes messages from numbers not saved in the user's contacts, looking for potentially suspicious patterns. The model is based on linguistic cues and conversation structure , learned from scams previously reported by other users.
If the system detects something suspicious, it displays a warning banner within the chat. This alert is private, meaning only the message recipient sees it, and the sender is unaware that their conversation has been flagged. From there, the user has several options: they can block the sender, report the chat to WhatsApp, continue the conversation if they believe it's legitimate, or mark the conversation as trusted. Marking the conversation as trusted removes the warning and prevents Scam Alert from flagging that chat again in the future.
One of the key advantages of this feature is that it doesn't require messages to leave the phone for analysis. Meta has emphasized that the content of conversations is neither sent to nor stored on external servers, thus maintaining end-to-end encryption. The company designed the tool based on three fundamental principles : the model and data always remain on the device, no automatic reports are generated about messages or alerts, and the user can disable the feature at any time in the settings.
Privacy at the heart of the debate
Privacy is undoubtedly one of the biggest concerns for users when it comes to artificial intelligence applied to messaging. That's why Meta has made it clear that Scam Alert does not read conversations. The analysis is performed locally, and the telemetry collected to evaluate the tool's performance is processed in confidential environments before being anonymized. The company has also published the model's weights in a public transparency log so that security researchers can audit its behavior.
In addition, users who mark a chat as trusted have the voluntary option to share their last five messages received via WhatsApp, with the aim of helping improve the system's accuracy. However, this only happens with the user's explicit consent. There are no automatic reports or unauthorized data sharing , which strikes an interesting balance between fraud protection and privacy.
The context: scams on WhatsApp are on the rise
The launch of Scam Alert is a response to a growing problem. According to data from the U.S. Federal Trade Commission (FTC), consumers reported losses of $425 million due to scams perpetrated through WhatsApp in 2025, a figure that is part of the more than $2.100 billion lost to fraud originating from social media that same year. Digital fraud knows no borders, and Spain is no exception.
In Spain, the Ministry of the Interior recorded 488.426 cybercrimes in 2025, a 5,1% increase over the previous year, and nearly nine out of ten were online scams. INCIBE (the National Cybersecurity Institute of Spain) has documented numerous cases in which criminals make contact via WhatsApp, move the conversation to Telegram, and offer small payments for liking videos or rating products before demanding increasingly larger sums. Fake job offers have become one of the most common lures , and many victims end up losing money after performing several seemingly harmless tasks.
Vacation-related scams are also common, with criminals impersonating hotels or booking platforms to steal personal and banking information. The Civil Guard and INCIBE (National Cybersecurity Institute) have repeatedly warned about this practice, which tends to intensify during the summer months. WhatsApp's new feature won't block these messages, but it will alert the user to pause and think before interacting with the sender.
What can users do to protect themselves?
While Scam Alerts are being rolled out to all users, experts recommend taking a few basic precautions to avoid falling into the scammers' trap. First, it's essential to activate WhatsApp's two-step verification, a measure that adds an extra layer of security to your account. It's also wise to be wary of unexpected links , especially if the message asks you to verify, connect, or link your account on the app.
Another important recommendation is to periodically check the devices linked to WhatsApp, in Settings > Linked devices, and log out of any that aren't recognized. Never follow instructions to link devices or scan QR codes unless you initiated the action yourself. And, of course, be very careful when a conversation that starts on Facebook or Instagram tries to migrate to WhatsApp—a classic tactic used by scammers to keep the conversation out of public scrutiny.
If you receive a suspicious job offer, it's best not to respond immediately. Before providing any information, verify if the company exists, if they have the job posting, and if the sender actually works there. Searching for snippets of the message online can help you find warnings from other victims , as scammers often reuse the same templates. And under no circumstances should you make any payments to secure a job or provide passwords, verification codes, or bank details.
If you have fallen for this scam, the first step is to contact your financial institution to try to block or recover the transaction. Afterward, it's advisable to gather all possible evidence and report the incident to the police. INCIBE offers the 017 helpline, a free and confidential cybersecurity support service for citizens and businesses, available 365 days a year from 09:00 a.m. to 21:00 p.m.
Scam Alert is presented as a promising tool in the fight against digital fraud, but it's not a magic bullet. The feature is optional, currently in beta, and doesn't yet have an official release date. In the meantime, the best defense remains caution and common sense. The basic rule with any unexpected offer is to check before you trust , and to be wary of anything that seems too good to be true. Technology advances, but so do scammers, and awareness remains the first line of defense.