Meta Platforms has introduced new AI-driven tools to detect and stop scams across Facebook, Messenger, and WhatsApp, aiming to curb a growing wave of online fraud that increasingly relies on social engineering and automation. The rollout marks an expansion of Meta’s ongoing efforts to use artificial intelligence for user safety across its apps.
On WhatsApp, the company has added alerts that warn users when suspicious device-linking requests are detected, a common method used by scammers to take over accounts by tricking individuals into sharing linking codes. Facebook is testing behavioral warnings for friend requests from accounts exhibiting high-risk signals, such as limited mutual connections or mismatched location data.
Messenger now features an AI-based alert that detects conversations resembling known scam patterns, including fraudulent job offers or financial solicitations. When triggered, users are prompted to have the messages reviewed by an automated system that can recommend blocking or reporting the sender.
Meta shared new data highlighting the scale of the problem. In 2025, it removed over 159 million scam advertisements, catching 92% proactively before user reports, and took down 10.9 million Facebook and Instagram accounts linked to organized scam networks. The company also participated in a multinational “Joint Disruption Week” in Bangkok, which led to over 150,000 accounts disabled and 21 arrests.
Meta said it plans to expand advertiser verification, with verified advertisers expected to account for 90% of ad revenue by 2026, up from 70% today. The company is under heightened scrutiny following reports that billions of higher-risk scam ads continue to circulate on its platforms, despite removals. European regulators also removed 5.9 million scam ads in 2025 under the EU’s Digital Services Act trusted flagger system.
