AI 詐騙警示功能
Meta 於旗下社交平台推出防詐騙功能
Meta 於 Facebook、WhatsApp 及 Messenger 上推出一系列由人工智能驅動的防詐騙功能，旨在更快速地揪出欺詐行為。這套系統會分析多項信號——包括文字、圖片及情境元數據——以標記潛在威脅，並配備名人冒充偵測及網域連結分析模組。
是次推出的具體保障措施包括：可偵測與公眾人物相關的虛假粉絲主頁及誤導性簡介的人工智能、自動移除連結至仿冒網站的內容、能顯示帳戶建立時間及共同好友的交友請求警示，以及顯示請求來源及操作紀錄的 WhatsApp 裝置連結提示。Messenger 用戶可透過「設定」>「私隱與安全」>「詐騙偵測」選擇退出此功能。
對消費者而言，這些工具旨在透過提供更清晰、實時的提示及自動移除機制，降低受騙風險，並提升面對可疑互動時的判斷能力。此舉反映出一個更廣泛的趨勢——大型社交平台正運用機器學習技術，將安全防護機制擴展至更大規模。
英文原文
Meta introduced a suite of AI-powered scam protection features across Facebook, WhatsApp and Messenger, designed to surface deceptive behavior more quickly. The systems analyze multiple signals — including text, images and contextual metadata — to flag potential threats, featuring modules for celebrity impersonation detection and domain-link analysis.
The rollout includes specific safeguards: AI that spots fake fan profiles and misleading bios tied to public figures, automated removal of content linking to cloned websites, friend-request warnings that surface account age and mutual connections, and WhatsApp device-linking alerts that show request origin and actions. Messenger users can opt out via Settings > Privacy & Safety > Scam Detection.
For consumers, these tools aim to reduce fraud risk by providing clearer, real-time cues and automatic removals, improving decision-making during suspicious interactions. The move reflects a broader trend of applying machine learning to scale safety controls on major social platforms.
- 來源
- Trend Hunter
- 發布
- 2026-03-26
- 品類
- Internet
- 出處
- zdnet, meta