May 13, 2026 · 10 min read

Meta Capped Its Own Anti Scam Spending at 0.15% of Revenue—Santa Clara County Just Sued Over $7 Billion in "Violating Revenue"

The first county level civil prosecution of its kind says Meta's own systems track 15 billion scam ads a day, charges scammers a premium to run flagged ones, and uses an internal "revenue guardrail" to keep cleanup costs from eating into the fraud revenue.

On May 11, Santa Clara County Counsel Tony LoPresti filed suit in Santa Clara County Superior Court accusing Meta Platforms of knowingly facilitating and profiting from billions of scam advertisements on Facebook and Instagram. The complaint, brought on behalf of all California residents, is the first lawsuit of its kind in California and the first by a local civil prosecutor anywhere in the United States targeting Meta's role in consumer ad fraud.

The headline number is $7 billion — Meta's annual revenue from ads that its own systems already flagged as likely fraud. The internal term for that money, per the complaint, is "violating revenue." The lawsuit's animating claim is not that Meta failed to detect scam ads. It is that Meta detected them, priced them, and decided how much to make from them.

"Meta is on the take," LoPresti said in a statement accompanying the filing. "Meta is taking very deliberate measures to make sure many of these scam ads get through."

A neoclassical courthouse in Santa Clara County at golden hour with the silhouette of a corporate office tower in the distance, suggesting a civil prosecution against a tech platform

The 0.15% Guardrail

The most damning passage in the complaint is not the $7 billion figure. It is a number that is six orders of magnitude smaller: 0.15%.

According to internal Meta documents previously surfaced by Reuters and incorporated into the Santa Clara complaint, Meta's antifraud team operates inside a "revenue guardrail." The guardrail is not a budget for catching scammers. It is the maximum slice of total company revenue that antifraud efforts are permitted to cost. That ceiling sits at 0.15% — roughly $135 million on a $200 billion top line.

If a proposed cleanup project is projected to take more than 0.15% of revenue off the table, the project gets throttled back. The cleanup function automatically yields to revenue when the two conflict. The filing alleges Meta booked $3.5 billion every six months from ads it had internally classified as carrying "higher legal risk" — the cohort the 0.15% rule is designed to protect.

This is the architecture detail compliance lawyers and journalists have been hunting for years. It is one thing to argue a platform was reckless with bad ads. It is another to point at a formula, written down in internal documents, that proves the platform priced the recklessness against revenue and chose revenue.

The Scale: 15 Billion Scam Ads a Day

The volume the complaint attributes to Meta's own internal tracking is hard to translate at first glance. Across Facebook, Instagram, and WhatsApp, Meta's systems flag approximately 15 billion scam advertisements every single day. Multiplied across a year, that is more than 5.4 trillion flagged scam impressions Meta serves while knowing what it is serving.

The volume gets converted into harm at the user end. The complaint references 2024 figures showing $2.5 billion in consumer fraud losses traceable to advertisements running on Meta's platforms. A separate Reuters investigation, central to the legal theory, found that Meta served ads were involved in roughly one out of every three successful internet scams against U.S. consumers in 2025.

Among the categories Santa Clara identifies in the filing:

Fraudulent financial products. Fake brokerage platforms, sham yield products, and impersonations of real fintech brands. Often paired with celebrity testimonial deepfakes.
Cryptocurrency schemes. Pig butchering pipelines, rug pull tokens, and counterfeit exchange landing pages. The complaint flags these as the highest loss category by dollar amount.
Celebrity impersonation. AI generated Elon Musk, Taylor Swift, and Warren Buffett endorsements asking for donations or routing to investment scams.
Military impersonation. Fake veterans' charities and active duty service member romance bait targeting older Americans.
Medical claims. "Cures" for terminal diseases, miracle weight loss supplements, and unapproved peptide products — explicitly aimed at users Meta's ad system has profiled as ill or aging.

The Premium Price for Flagged Ads

The mechanism that elevates this from negligence to civil prosecution is the price Meta allegedly charges scammers once their ads have been flagged.

When Meta's classifier scores an advertiser as likely fraudulent, the lawsuit alleges the platform does not block the ad. It marks up the price. The premium is presented internally as a "risk surcharge" — additional revenue per impression collected from advertisers Meta's own systems suspect are scamming users. The economic effect is that the worse a scammer's reputation inside Meta's risk model, the more they pay per click, and the more Meta earns from continuing to serve their ads.

The complaint also alleges Meta operates a tier of "trusted business partners" who are allowed to vouch for scam adjacent advertisers and effectively launder them through a higher trust pipeline. And it cites instances where the same scam advertiser was flagged 500 or more separate times — by Meta's own automated systems — and was still permitted to keep buying inventory.

The targeting layer is what closes the loop. Meta's ad platform exposes audience segments to advertisers based on inferred age, financial stress, recent searches, and inferred medical conditions. The complaint argues that those targeting affordances are made available to advertisers Meta's risk model has flagged as suspect — meaning a known bad advertiser can hand pick the most vulnerable user cohort in the country and pay Meta extra for the privilege.

Why a County Counsel

Federal cases against social platforms have historically died on Section 230, the 1996 statute that immunizes platforms from liability for user content. Santa Clara's filing is engineered to sidestep that defense.

The complaint is not framed as a third party content claim. It is framed as a first party commercial conduct claim under California's Unfair Competition Law and False Advertising Law. The theory: Meta is not being sued for what scammers said on its platform. Meta is being sued for what Meta itself did with its own ad system — selling premium placement to advertisers it had already flagged as fraudulent, capping its own enforcement budget below the level needed to stop them, and targeting victims through proprietary tools that had nothing to do with user generated content.

That framing puts the case in the same legal lane as recent state actions like the California Privacy Protection Agency's enforcement push and the $12.75 million GM OnStar settlement Attorney General Bonta announced last week. Both treat a company's own commercial design as the wrongful act, not the conduct of third parties using its platform.

Santa Clara is asking the court for injunctive relief, restitution to defrauded consumers, and civil penalties of up to $2,500 per violation — with enhanced penalties available for violations targeting senior citizens. If the court accepts the per impression theory of "violation," the math gets large quickly.

Meta's Response

Meta says the lawsuit is built on a Reuters report that "distorts our motives and ignores the full range of actions we take to combat scams every day." The company says it removed 159 million scam ads in 2025 and that it "aggressively fights scams" through both detection systems and partnerships with law enforcement.

The 159 million number is large in isolation. Against the 15 billion per day figure in Meta's own internal tracking, it is roughly 2.9% of the flagged ad inventory for a single calendar year. The complaint's argument is that the gap between flagged and removed is not a detection failure. It is a deliberate revenue protection choice locked in by the 0.15% guardrail.

Senators Richard Blumenthal and Josh Hawley separately called this week for a federal investigation of Meta's role in consumer scams, citing the same Reuters reporting Santa Clara's complaint relies on. The federal track and the state track are now both alive at the same time.

Why This Matters Beyond the Ad System

The pattern Santa Clara is naming — automated flagging followed by deliberate non action because the flagged thing is profitable — is not unique to Meta's ad system. It is the operating model of almost every large platform that monetizes user attention.

Email is the closest parallel. Marketing email senders deploy authenticated infrastructure (DKIM, DMARC, dedicated IPs) to make sure their messages reach the inbox. Major email sending platforms have automated systems that detect when a customer's mailing list contains scraped or fraudulent addresses, when their content trips spam classifiers, or when their bounce rate signals an unsolicited campaign. Those signals are routinely logged and routinely ignored, because suspending a paying customer's account costs real revenue.

The same logic applies to tracking pixels. Senders embed pixels to learn who opened, when, from where, and on what device. Email service providers know how the pixels work, know which senders use them aggressively against unsubscribed recipients, and know which campaigns are functionally a wiretap on the recipient's reading habits. None of it gets blocked, because deliverability fees are the product. The recipient has no equivalent of a Tony LoPresti walking into court on their behalf.

That same pixel mechanism is what produced the $10 million Forbes CIPA settlement last week, the only difference being that web trackers have started attracting plaintiff lawyers while email trackers still mostly don't. The Santa Clara theory — that platforms are liable for their own commercial design, not for third party content — is portable. If the court accepts it, the ground shifts for every business model that depends on monetizing flagged user behavior.

What Compliance Officers Should Watch For

If you sit in a privacy, trust and safety, or compliance role at a platform company, three signals from the Santa Clara filing matter immediately:

Internal classifications survive discovery. Meta's own "violating revenue" label became Exhibit A. If your trust and safety team labels accounts, advertisers, or campaigns with risk tiers, those labels are discoverable and they constitute knowledge. The label is also the wording prosecutors will quote.
Revenue ceilings on enforcement budgets are a litigation hazard. The 0.15% figure is the document that turns a regulatory complaint into a populist one. Any policy that pegs enforcement spending to a fraction of revenue is now subpoena bait.
First party design beats Section 230 immunity. The defense that worked for a decade is being routed around. Plaintiff lawyers will increasingly frame platform decisions as commercial acts rather than content moderation, especially where those decisions involve pricing, targeting, or partner programs.

For everyone else, the takeaway is that the largest social platform in the world allegedly built a formula to decide how much fraud it would tolerate as a function of revenue — and the formula came out to "almost all of it." A county counsel one bridge away from Menlo Park has decided that is a state law violation. The next 18 months of discovery will determine whether the formula generalizes to every other platform that runs the same playbook.