Meta Enhances Security on WhatsApp and Messenger with AI

Meta Deploys AI Tools to Secure Its Messaging Services

On March 11, 2026, Meta announced the introduction of new AI-based detection tools on its platforms Facebook, WhatsApp, and Messenger. These systems are designed to identify suspicious behaviors in order to prevent the increasing scams that are proliferating on these services.

While the idea of having private conversations monitored may be concerning, Meta claims that this initiative aims to protect users against a surge in online scam attempts that target millions of people daily.

Protecting Conversations on WhatsApp

Meta does not directly read messages exchanged on WhatsApp, thanks to end-to-end encryption that ensures only the sender and recipient can decrypt them. However, the company analyzes connection signals to detect unusual behaviors, particularly when linking a new device.

According to a post published on Meta's blog, these measures are intended to counter sophisticated scams based on social engineering. For example, a fraudster might attempt to link an account to their device by using deceptive scenarios to obtain sensitive information.

The most common scenario currently is the fake talent contest. A scammer contacts you pretending to be an acquaintance and asks you to vote for them on an external site. Once you enter your phone number there, the scammer attempts to link your account to their own computer.

WhatsApp is now capable of detecting connection requests from unusual sources, such as an inconsistent location, and alerts the user of a possible fraud attempt.

Facebook and Suspicious Friend Requests

In parallel, Meta is testing an alert system on Facebook for friend requests deemed suspicious. This system analyzes geographical consistency and the number of mutual friends to identify malicious outreach attempts.

Users receive an alert if a request comes from an account with few or no mutual friends, or from an inconsistent geographical region. While AI provides warnings, the final decision to accept or decline the request remains with the user.

The analysis also focuses on the geographical location of the sender. If a request comes from a region that is completely inconsistent with your usual network or recent activities, Facebook considers this a red flag.

The goal is to provide an explicit warning even before you open the relevant profile. This way, you can take the time to review the request before deciding.

However, note that Meta has chosen not to automatically block these requests. The final decision is in your hands. In any case, AI is not infallible. It may sometimes flag someone you have just met in a specific context.

Extending Security Measures on Messenger

Meta is also extending its scam detection system to Messenger in several countries. When a conversation with a new contact shows signs of fraud, such as suspicious job offers, the user is alerted.

Messenger then offers to share certain messages for analysis. If a scam attempt is detected, the app informs the user about common frauds and suggests actions such as blocking the suspicious account.

A Continuous Effort Against Scams

Meta is tackling a persistent problem on its platforms. According to Reuters, fake advertisements generate about $16 billion annually for Meta. The company removed over 159 million fraudulent ads last year, with 92% being taken down before users reported them.

Additionally, Meta has dismantled numerous scam networks, removing 10.9 million accounts linked to these activities on Facebook and Instagram. These efforts aim to make the platforms safer for all users.

In short, Meta's goal is to make its platforms safer for both individuals and businesses. So, will these new features be enough to reassure users?