近日，有關Meta公司廣泛依靠高風險及欺詐性廣告獲利的內部文件曝光，引起全球媒體與監管機構高度關注。根據文件顯示，Meta在2024年的總收入中約有10%（約160億美元）來自欺詐及違禁商品廣告，其旗下社交平台Facebook、Instagram和WhatsApp每天展示高風險廣告高達150億條。這些欺詐性廣告涵蓋範圍廣泛，包括虛假電商促銷、投資陷阱及高收益理財騙局、非法線上博彩，以及違禁醫療產品如偽劣藥品等。此舉導致全球數十億用戶長期暴露於高風險內容之中，嚴重影響平台信任度。

文件顯示，Meta至少在過去三年中未能有效識別並阻止大量欺詐廣告，即使系統發現潛在欺詐行為，多數情況下也僅在自動化系統判定欺詐概率達95%以上時才予以封禁；對於低於這一概率但仍存在風險的廣告，Meta則採取提高廣告費率的方式“懲罰”，形成一種“明禁暗放”的操作模式。更為關鍵的是，Meta利用AI和大數據技術，將曾點擊過欺詐廣告的用戶標記為「易感人群」，不斷推送類似內容，形成所謂的“欺詐閉環”。

該模式的後果顯著。報告指出，美國約三分之一的詐騙案件與Meta產品相關；英國2023年支付類詐騙損失中有54%涉及其平台；新加坡2025年上半年82%的線上詐騙案件與Meta有關，冒充政府進行的詐騙更激增200%。業內人士分析，Meta之所以對欺詐廣告採取“明禁暗放”的策略，與違法成本與收益不平衡有關。據估算，監管罰款上限僅10億美元，遠低於Meta半年內來自高法律風險欺詐廣告的35億美元收入。文件還指出，Meta的產品已成為全球“欺詐經濟”的重要支柱。

對於此類指控，Meta發言人安迪斯通回應稱，文件的視角“片面”，10.1%的收入占比是“粗略且過度包容的估算”，實際比例更低，因其中包含大量合法廣告，但未提供具體數據。此事件再次將Meta的廣告審查機制與企業道德責任推向輿論風口，並引發全球監管機構對其業務模式和風險管理的進一步關注。

Recently, internal documents regarding Meta’s heavy reliance on high-risk and fraudulent advertising have been leaked, drawing global media and regulatory attention. According to the documents, approximately 10% of Meta’s 2024 revenue—around $16 billion—came from fraudulent and prohibited product ads. On its platforms—Facebook, Instagram, and WhatsApp—Meta displayed up to 15 billion high-risk ads per day. These ads covered a wide range of scams, including fake e-commerce promotions, investment traps and high-yield financial schemes, illegal online gambling, and prohibited medical products such as counterfeit medicines. As a result, billions of users worldwide were continuously exposed to high-risk content, seriously undermining trust in the platforms.

The documents also revealed that Meta failed to identify and block a large volume of fraudulent ads over at least the past three years. Even when potential fraud was detected, the company reportedly only banned advertisers if its automated system predicted a fraud probability of 95% or higher. For cases deemed lower risk but still suspicious, Meta applied a “penalty” by increasing advertising rates rather than blocking the ads—creating a so-called “allow but penalize” approach. Additionally, Meta reportedly leveraged AI and big data to mark users who had previously clicked on fraudulent ads as “vulnerable” and continued to target them with similar content, creating a self-reinforcing “fraud loop.”

The consequences have been significant. Reports indicate that roughly one-third of U.S. fraud cases are linked to Meta’s products, 54% of payment-related fraud losses in the U.K. in 2023 involved its platforms, and in Singapore during the first half of 2025, 82% of online fraud cases were connected to Meta, with government impersonation scams surging by 200%. Industry analysts have noted that Meta’s “allow but penalize” strategy may be driven by the imbalance between potential penalties and revenue from risky ads—regulatory fines are capped at around $1 billion, far lower than the $3.5 billion generated from high-legal-risk fraudulent ads in just six months. The documents further indicate that Meta’s platforms have become a central pillar of the global “fraud economy.”

In response to the allegations, Meta spokesperson Andy Stone stated that the documents provide a “one-sided perspective” and that the 10.1% revenue figure is a “rough and overly inclusive estimate,” with the actual proportion being lower due to the inclusion of many legitimate ads. No specific data was provided. This revelation has reignited global scrutiny of Meta’s advertising oversight and corporate responsibility, highlighting ongoing concerns about its business model and risk management practices.