Meta projected in late 2023 that about 10% of its total annual revenue—nearly $16 billion—would come from ads related to scams and banned items, according to internal documents. Previously unreported documents reviewed by Reuters reveal that for at least three years, Meta failed to identify and stop a surge of fraudulent ads, exposing users on Facebook, Instagram, and WhatsApp to scams, illegal gambling, and banned medical products.
An average document from December 2024 estimates that Meta shows about 15 billion “high-risk” scam ads daily. These ads exhibit clear signs of fraud, earning Meta approximately $7 billion annually, as noted in another late 2024 document.
Many fraudulent ads came from suspicious marketers flagged by Meta’s internal systems. However, advertisers are only banned if the system predicts a 95% certainty of fraud. If less certain, Meta increases ad rates to deter suspect advertisers, according to the documents.
Users who click on scam ads are likely to see more due to Meta’s ad-personalization system, which tailors ads based on user interests.
The documents, spanning 2021 to the current year, reflect Meta’s efforts to quantify abuse on its platforms and its reluctance to take actions that might harm business interests.
Meta’s acceptance of revenue from suspected fraudulent sources underscores a lack of regulatory oversight in the advertising industry, said Sandeep Abraham, a former Meta safety investigator and current fraud examiner at Risky Business Solutions.
“If regulators wouldn’t tolerate banks profiting from fraud, they shouldn’t allow it in tech,” he told Reuters.
Meta spokesman Andy Stone stated that the documents show a selective view that distorts Meta’s fraud approach. The initial estimate of earning 10.1% of 2024 revenue from scams was “rough and overly-inclusive,” as it included many legitimate ads. He declined to provide updated figures.
“The assessment was to validate planned integrity investments, including combating fraud, which we did,” Stone said. “We fight scams aggressively as neither users, advertisers, nor we want this content.”
Stone mentioned that scam ad reports were reduced by 58% globally over 18 months, and more than 134 million scam ads were removed in 2025.
Some documents show Meta’s commitment to reducing ad scams, aiming for a 50% reduction in certain markets by 2025. Managers praised staff for successful scam reduction efforts.
Meta’s research suggests its products are integral to the global fraud economy. A May 2025 safety presentation estimated that one-third of all successful U.S. scams involved Meta platforms. Other documents acknowledged that competitors are better at preventing fraud.
An internal review in April 2025 concluded that advertising scams on Meta are easier than on Google. The document didn’t elaborate on the reasons.
Regulators worldwide are pressing Meta to improve its user protection from online fraud. The U.S. SEC is investigating Meta for running ads for financial scams, and a UK regulator found Meta’s products involved in 54% of scam losses in 2023.
The SEC and UK regulator didn’t respond to questions. Meta’s Stone pointed to SEC disclosures stating that efforts to address illicit advertising would affect revenue.
Amid regulatory pressure, Meta is investing heavily in AI, planning up to $72 billion in capital expenditures this year. CEO Mark Zuckerberg assured investors that Meta’s ad business could support this investment.
“We have the capital from our business to do this,” he said, announcing a new data center in Ohio.
Meta is evaluating the cost of enhancing scam ad enforcement against potential government penalties. Documents show Meta’s goal to reduce illicit revenue while considering business impact.
Meta plans to cut scam ad revenue on Facebook and Instagram but anticipates regulatory fines up to $1 billion. Yet, fines would be much smaller than scam ad revenue, according to a separate November 2024 document.
Rather than proactively addressing advertiser vetting, Meta’s leadership decided to act only in response to regulatory action.
Stone disputed claims that Meta would only act if forced, asserting it isn’t company policy.
Meta has set limits on revenue loss from actions against suspect advertisers. In early 2025, the team handling advertiser vetting couldn’t take actions costing more than 0.15% of total revenue, around $135 million of the $90 billion generated in the first half of 2025.
“Let’s be cautious,” the manager noted, referring to revenue guardrails.
Stone clarified that the 0.15% figure was from a projection, not a hard limit.
Under increasing pressure, Meta executives presented Zuckerberg with a moderate enforcement plan in October 2024, targeting countries facing regulatory action. The goal was to reduce scam-related revenue from 10.1% in 2024 to 7.3% by 2025, and to 5.8% by 2027.
A 2022 document detailed a network of accounts impersonating U.S. military members to defraud Facebook users. Sextortion and fake accounts impersonating celebrities or brands were also prevalent.
Despite rising fraud, a 2022 document noted Meta’s “lack of investment” in automated scam detection, classifying scam ads as a “low severity” issue.
Layoffs hindered enforcement, with resources shifted to virtual reality and AI, limiting safety staff’s computing resources use.
Stone said staffing for scam ad addressing had substantially increased.
Meta ignored most user scam reports, with safety staff estimating 100,000 valid weekly reports in 2023, but 96% were ignored or wrongly rejected.
Meta aimed to reduce ignored reports to no more than 75% in the future.
Erin West, a former prosecutor, noted Meta’s default response was to ignore user fraud reports.
Last October, a Canadian Air Force recruiter was hacked and her account used for a crypto scam. Despite multiple reports, Meta didn’t respond for weeks, causing more victims. Police traced funds to Nigeria but recovering them is unlikely.
Meta declined to comment on the hacked account.
Internally, Meta refers to such scams as “organic,” including fraudulent classified ads, fake dating profiles, and bogus medical cures.
Meta’s platforms are exposed to 22 billion organic scam attempts daily, in addition to 15 billion scam ads.
Meta’s fraud policing misses much scam activity. In Singapore, only 23% of police-reported scams violated platform policies.
Some staff noted gaps in scam policies, like $250,000 in scam crypto ads from a fake account claiming to be Canada’s prime minister.
Meta’s lenient rules allow small advertisers to commit multiple frauds before being blocked, while “High Value Accounts” could accrue over 500 strikes without shutdown.
Fraud campaigns can be extensive, with four removed ads generating $67 million monthly.
An employee’s “Scammiest Scammer” report highlighted top complaint earners, but many accounts remained active.
Meta began charging suspected fraudsters more to reduce scam ads, offsetting revenue loss. Penalty bids aimed to make suspicious advertisers less competitive.
Stone said penalty bids reduced scam reports and slightly decreased overall ad revenue.
