Facebook banned roughly 3.5 billion fake accounts across its platform in 2025 and more than 38 billion over the past eight years, according to an analysis published today by VAB, the video advertising trade body. The figures, drawn from Meta's own transparency data, arrive as the company faces mounting legal and regulatory pressure over the scale of fraud and scam advertising flowing through Facebook and Instagram.

The report, titled "Friend or Frenemy?", examines the reasons why Facebook removes billions of accounts each year and what those removals signal about brand safety for advertisers. Its central question is blunt: given the persistence of fake accounts, scams, and illicit goods across Meta Platforms, is the company a friend or a frenemy to the marketers who fund it.

VAB built its account-ban figures from the Meta Transparency Center, specifically the "Accounts Actioned" metric under Fake Accounts for Facebook, referenced as of April 28, 2026. The organization notes an important gap in the underlying disclosure. According to VAB, Meta has not published the number of false and duplicate accounts within its SEC-regulated filings since 2023, leaving the transparency center as the primary public source for this data.

Eight years of bans, measured against the world's population

The year-by-year totals show considerable volatility rather than a steady climb. Facebook actioned 3.34 billion fake accounts in 2018, then 6.50 billion in 2019, 5.80 billion in 2020, 6.50 billion again in 2021, and 5.80 billion in 2022. The count dropped sharply to 2.62 billion in 2023 before rising to 4.33 billion in 2024 and settling at 3.50 billion in 2025.

To convey the magnitude, VAB compared each year's ban count against global population estimates sourced from the United Nations Department of Economic and Social Affairs, Population Division, using the 2024 revision of World Population Prospects. The comparison produces figures that are difficult to contextualize. In 2019, the 6.50 billion accounts banned represented 83 percent of a global population of 7.81 billion. In 2025, the 3.50 billion figure equalled 43 percent of an 8.23 billion world population. VAB frames the aggregate as equivalent to nearly half the world's population.

A second comparison sharpens the point. VAB measured the ratio of fake accounts banned to Facebook's monthly active users, drawn from Meta's annual company filings and based on the average of quarterly worldwide figures. In several years, Facebook banned more fake accounts than it counted actual monthly active users. The ratio reached 2.7 in 2019, when 6.50 billion accounts were banned against 2.43 billion monthly active users, and stood at or above 2.0 through 2022. That relationship cannot be extended into recent years, however, because Meta stopped reporting monthly active Facebook users in its disclosures. The metric is marked "No Longer Reported by Meta" for both 2024 and 2025.

What a banned account actually represents

VAB is careful to frame the ban totals as a proxy rather than a direct measure of harm. The number of banned accounts, the analysis states, is indicative of the scale of consumer and brand safety issues involving fraud, scams, and the selling of illicit goods across Meta Platforms. Fake accounts undermine consumer trust and create brand safety risks for advertisers, the report argues, with perpetrators using false identities to harm consumers and drive deceitful actions through bots.

That framing matters for how the data should be read. A high ban count can indicate aggressive enforcement, a large underlying problem, or both. VAB, whose members produce professionally produced video content positioned as an alternative to social platforms, has a commercial interest in highlighting brand safety weaknesses on Meta's properties. The report's closing section points marketers toward premium video as a route to what it calls real performance in a brand-safe environment. Readers weighing the findings should hold that context alongside the underlying transparency figures, which originate with Meta rather than with VAB.

The scam advertising backdrop

The account-ban figures do not exist in isolation. VAB layers in a series of findings, drawn from external reporting and Meta's own statements, that describe an advertising environment saturated with fraud.

The most consequential of these traces to a Reuters investigation published on November 6, 2025, built on internal Meta documents. That reporting, which PPC Land covered when Meta charged suspected fraudsters premium rates while earning billions from scam ads, indicated that Meta internally projected roughly 10 percent of its 2024 revenue, an estimated 16 billion dollars, would come from advertisements promoting scams and banned goods. The same documents estimated that Meta platforms delivered over 15 billion higher-risk scam advertisements each day in 2024.

VAB reproduces two direct quotations from the source material. One, attributed to a Meta internal review cited by Reuters, states: "It is easier to advertise scams on Meta platforms than Google." The other, attributed to the Meta Newsroom and dated December 3, 2025, reads: "Scams don't just harm individual victims, they undermine trust in our entire advertising ecosystem, which is the very foundation of our business model."

Meta's own enforcement disclosures form part of the picture. According to a Meta statement dated December 3, 2025, the company's detection systems identified and disrupted nearly 12 million accounts across Facebook, Instagram, and WhatsApp in the first half of 2025 that were associated with criminal scam centers. PPC Land reported at the time that Meta removed 134 million scam ads in 2025 and that user reports about scam ads had declined more than 50 percent over the preceding 15 months, figures the company presented at the Global Anti-Scam Summit in Washington.

New advertisers and impersonation

Two further data points, sourced to New York Times reporting from May 15, 2025, describe how the fraud manifests at the advertiser level. VAB cites a figure that 70 percent of newly active advertisers on Instagram and Facebook were promoting scams, poor-quality products, or illicit goods. Separately, the analysis references an instance in which 4,400 ads were found impersonating a real business's name and phone number across the two platforms, while the legitimate business itself had purchased only 15 ads.

The impersonation problem is one PPC Land has documented in concrete terms. In coverage of how Meta sued scam advertisers in Brazil, China, and Vietnam, the site detailed the celebrity-impersonation and cloaking techniques used to redirect users to fraudulent websites, alongside the cease-and-desist letters Meta issued to marketing consultants.

The strike threshold that defines the ceiling

Perhaps the most pointed argument in the VAB analysis concerns how many violations an advertiser can accumulate before facing a ban. Citing the New York Times reporting, VAB states that Meta advertisers are allowed between 8 and 32 financial fraud strikes before their account is banned. Strikes, the report explains, are a punitive, step-by-step enforcement system applied when ads or connected pages violate Community Standards or Advertising Policies.

VAB draws a logical inference from that threshold. Because Meta permits many violations before acting, the analysis argues, it is reasonable to believe Facebook should have banned far more than 3.5 billion accounts. The high tolerance for repeated violations, in other words, suggests the reported ban totals understate the true scale of problematic activity on the platform.

That tolerance connects to a broader debate PPC Land has tracked. In coverage of expert reaction, captured in "This is not the sword I'd die on", advertising technology educator Nicole Pruess questioned whether the 95 percent certainty threshold Meta applies before banning an advertiser could be lowered without sweeping legitimate businesses into overzealous fraud detection. The strike system and the certainty threshold together describe an enforcement posture calibrated to avoid false positives, which necessarily tolerates a degree of genuine fraud.

Why the numbers matter for marketers

For advertisers and agencies, the significance lies less in any single statistic than in what the accumulation describes: a platform environment where the same ad delivery infrastructure serving legitimate brands also carries billions of daily scam impressions. When a large share of scam-related losses flows through that shared infrastructure, it affects consumer trust in advertising broadly, a dynamic PPC Land has traced across multiple investigations.

The regulatory and legal consequences are no longer hypothetical. The Consumer Federation of America filed a class action lawsuit against Meta in April 2026, an action PPC Land examined when a consumer group sued Meta over scam ads that fund billions in revenue. That complaint drew directly on the Reuters documents, alleging Meta ignored or incorrectly rejected the overwhelming majority of user scam reports and set enforcement thresholds tilted in favor of high-spending advertisers.

Enforcement pressure is also reshaping the compliance burden on legitimate advertisers. Meta has steadily expanded identity-verification requirements for higher-risk categories, moves PPC Land covered as the company mandated SEBI verification for India securities ads and expanded advertiser verification for Thailand campaigns. The pattern extends to transparency obligations: PPC Land reported that Meta will force ad spend disclosure to advertisers starting February 2027, granting end advertisers a formal right to see how intermediaries configured and spent against their budgets.

The financial stakes for Meta remain substantial. The company's advertising business generated 58.1 billion dollars in the fourth quarter of 2025 alone, with full-year 2025 advertising revenue reaching 196.2 billion dollars. Against that backdrop, the projected 16 billion dollars tied to scam and banned-goods advertising represents a meaningful share of the total, which is precisely why enforcement decisions carry commercial weight for the platform itself.

VAB's contribution to this record is narrower but distinct. By assembling eight years of ban data into a single view and setting it against global population and monthly active user benchmarks, the analysis translates an abstract enforcement metric into a scale that advertisers can weigh. Whether that scale reads as evidence of a platform cleaning up its ecosystem or of a problem too large to contain depends heavily on the strike thresholds, the certainty settings, and the revenue incentives that surround every one of those bans.

Timeline

Summary

Who: VAB, the video advertising trade body, published the analysis, which draws on Meta's own transparency data and external reporting from Reuters and the New York Times. Meta Platforms is the subject of the findings.

What: The report, "Friend or Frenemy?", documents that Facebook banned roughly 3.5 billion fake accounts in 2025 and more than 38 billion over eight years, sets those totals against global population and monthly active user benchmarks, and situates them within a broader record of scam advertising, including a projected 16 billion dollars in 2024 scam-ad revenue and a strike system allowing 8 to 32 fraud violations before an account ban.

When: VAB published the analysis on July 20, 2026, referencing Meta Transparency Center data as of April 28, 2026.

Where: The findings concern Facebook and Instagram specifically and Meta Platforms more broadly, with population data sourced from the United Nations and enforcement context spanning Meta's global operations.

Why: The account-ban totals serve as a proxy for the scale of fraud, scams, and illicit-goods activity on Meta's platforms, a matter of direct concern to advertisers weighing brand safety risk against the reach of the company's advertising infrastructure.