Meta Uses AI to Combat Scams on Its Platforms

Meta recently unveiled a new series of AI-based tools designed to combat the scams that proliferate on its platforms such as WhatsApp, Facebook, and Messenger. This initiative comes as the number of frauds, often orchestrated via fake celebrity profiles, continues to rise. In 2025, Meta has already removed over 159 million fraudulent ads, highlighting the scale of the problem.

A spokesperson for Meta did not immediately respond to requests for comments on this announcement. Recent scams heavily exploit AI-generated content to create fake profiles and even fraudulent websites. It makes sense that Meta would use its own AI tools to protect its users from these threats.

Warnings on WhatsApp

WhatsApp, as a global messaging app, is a frequent target for scammers, particularly due to its mobile device linking features that allow for quick connections of external devices to an account. Once a scammer gains access to an account, they can impersonate the user to deceive their friends and family. Last year, Meta removed nearly 7 million accounts used by scammers in this manner.

Scammers often pose as relatives, sellers, service providers, or content creators, asking for phone numbers to share device linking codes and QR codes that some people might activate without thinking. The new WhatsApp feature will monitor for signs of danger from unlinked accounts, such as a request coming from a country with which you have never communicated before. It will then display an additional pop-up to ensure that you want to connect with a stranger, ideally giving you a moment to reflect on the appropriateness of this decision.

Warning Signals on Facebook and Messenger

On Facebook, alert signals are in place to identify suspicious accounts, particularly those that have just been created. Meta is also adding new protections against scams on Facebook, including warnings about adding or communicating with a suspicious account. Although Meta has not revealed all the signs it uses to determine if an account is suspicious, like WhatsApp, it looks for unusual locations and those outside the country.

Another alert signal is an account with which you or your friends have never interacted before. Facebook will include explanations next to its warning pop-ups, such as noting that an account was created only a few days ago, before you confirm your choice. The Messenger app is introducing a similar feature that monitors account activity to detect behaviors indicative of scams. Meta specifically mentioned job-related scams, where scammers offer impossible or highly paid jobs to steal personal information or direct people to a phishing site.

AI Scans to Spot Fraudulent Content

Messenger now includes AI scans to monitor for signs of scamming, but it is unclear how extensively this is being done. Regarding phishing sites, their number is on the rise, and they appear more authentic than ever, even deceiving those experienced in the dangers of phishing. A key example of generative AI is the ability to easily create a fraudulent web page and mimic real brand logos. Meta reports that it uses its own AI tools to scan and more accurately identify fraudulent content. This includes reviewing Messenger histories, monitoring fake photos impersonating celebrities, and detecting unusual mentions of public figures. These tools can also detect phishing patterns on the site.

Meta's new AI tools for detecting scams naturally raise privacy concerns. While the company claims that the features analyze account activity to spot fraudulent behaviors, it has not specified how this data is processed or stored. Past controversies surrounding Meta's AI initiatives, including its augmented reality glasses, show that AI-based surveillance can carry unforeseen privacy risks, underscoring the importance of transparency and robust protections when deploying such technologies on personal messaging platforms.

Meta has not specified when these tools will be available to all users.