AI detects impersonation fraud
According to an announcement on March 11 (local time), Meta said that scammers are using more sophisticated tricks, including impersonating celebrities, brands or reputable organizations to deceive users.
This forces the company to expand the application of AI and strengthen coordination with law enforcement agencies to detect and prevent fraud acts early.
Meta said the new AI systems are designed to detect scams that take advantage of celebrity images or impersonate brands.
The system will analyze many different factors such as text content, images, account behavior and related context.
Through the synthesis of data signals, AI can identify suspicious signs such as fake fan expressions, misleading biography information, or unusual connections with public figures.
This technology also helps detect fraudulent links and fake domain names. These websites are often designed to resemble legitimate platforms to lure users to provide personal information or financial data.
According to Meta, the new system helps detect more fraudulent acts than before, while contributing to protecting thousands of brands from the risk of being counterfeited in the online environment.
Increase safety warnings on applications
Along with server-side detection tools, Meta is also testing many direct safety warnings on its applications.
On Facebook, the company is testing warnings for suspicious friend requests. If the account sending the invitation shows unusual signs, such as very few mutual friends or writing a country other than the actual location, the system will display a notification for users to consider before accepting.
Meanwhile, WhatsApp will add a warning to prevent account hijackings.
A common trick today is to trick users into entering authentication codes or scanning QR codes to link accounts to stolen devices. The new system will detect suspicious requests and promptly warn.
On Facebook Messenger, Meta also expanded the AI fraud detection feature to more countries.
This tool can flag popular scam patterns such as suspicious job offers or fake investment proposals, while allowing users to send the nearest conversation for AI to analyze before deciding to block or report accounts.
Tighten advertising and law enforcement cooperation
In addition to detecting fraud, Meta is also expanding the advertiser verification process. The company aims that by the end of 2026, verified advertisers will account for about 90% of advertising revenue, up from the current 70%.
Meta emphasizes that cooperation with law enforcement agencies is still an important part of the anti-fraud strategy.
In 2025, the company removed more than 159 million scam ads on its platforms, of which 92% were detected before users reported.
Meta also deleted 10.9 million accounts on Facebook and Instagram related to fraud centers.
