A series of internal Meta documents recently disclosed by Reuters has shed light on a controversial corner of the tech giant's global advertising ecosystem. The leaked files reveal that in 2024, approximately 10% of Meta's total revenue—amounting to roughly $16 billion—was generated from high-risk, deceptive, or prohibited advertisements.

The Mechanics of High-Risk Ad Delivery

According to the leaked documents, Meta's platforms serve approximately 15 billion high-risk ads to users daily. While many of these ads are flagged by internal systems as "suspected fraud," the platform's automated enforcement protocols reportedly only mandate a forced takedown when system confidence in the violation reaches 95% or higher.

This threshold creates a gray area: as long as there is a margin of uncertainty, these ads continue to run. In some cases, lower-trust advertisers are permitted to keep their campaigns active by paying higher acquisition costs. The documents explicitly note that completely eliminating these high-risk ads would have a substantial negative impact on Meta's overall revenue, highlighting an ongoing internal tension between commercial interests and policy compliance.

How the Ad Auction System Capitalizes on Risk

This dynamic is deeply embedded in how modern digital ad auctions operate. Meta's system dynamically adjusts bidding requirements based on advertiser trust scores. Instead of facing immediate bans, lower-trust advertisers often face higher pricing, essentially paying a premium for exposure. Meanwhile, machine learning algorithms optimize delivery based on user interactions: if a user clicks on a suspicious ad, the system is trained to serve them similar content, creating a feedback loop that drives both engagement and platform revenue.

Advertising accounts for over 97% of Meta's total revenue. When a significant portion of this income is tied to high-risk content, it introduces long-term risks regarding user safety, advertiser trust, and regulatory scrutiny.

The Revenue Red Line in Ad Moderation

The leaks also point to specific operational constraints placed on Meta's safety and moderation teams:

  • The 0.15% Revenue Cap: A 2024 internal memo revealed that moderation teams were instructed not to let ad enforcement actions cause more than a 0.15% drop in overall ad revenue.
  • User Report Backlog: A separate internal report from 2023 indicated that Meta rejected or took no action on approximately 96% of user-submitted scam reports.

The Impact on Legitimate E-commerce Advertisers

In response to the reports, Meta stated that the leaked documents were taken out of context and that the 10% figure was a broad estimate that included legitimate ad categories. The company also emphasized that it had removed over 134 million scam ads over the past 18 months.

For legitimate e-commerce brands and performance marketers, this aggressive automated cleanup has a familiar downside. Many compliant advertisers frequently experience accidental ad disapprovals, sudden account flags, and unexpected bans as Meta's automated systems attempt to sweep up bad actors, often catching legitimate businesses in the crossfire.

A Systemic Industry Challenge

This challenge is not unique to Meta. As AI-driven, automated advertising systems become the industry standard across major platforms, detecting sophisticated "risk ads" has become highly complex. Bad actors constantly rotate domains, cloaking pages, and creative assets to bypass automated reviews. Because these high-risk ads often yield high click-through rates, they generate substantial revenue, creating a complex balancing act for platforms trying to manage growth alongside ecosystem health.