Security1 distinct publisher3 min readPublished
Documents reviewed by Reuters put Meta's projected 2024 scam and banned-goods ad revenue near $16 billion, while a February 2025 internal plan set aside $135 million in forgone revenue to curb suspicious advertisers.
The Watch · Security desk
Compiled by The WatchSomething wrong?How this is made
Run the two revenue figures against each other and they do not share a denominator. If about $16 billion is 10 percent of the total, the base is about $160 billion [1][1]. Take 0.15 percent of that and you get roughly $240 million, above the $135 million the February 2025 document names [8][2]; back-solve instead from $135 million at 0.15 percent and the implied base sits nearer $90 billion [3]. Reuters reviewed the cache, and we are working from its reporting rather than the scoping footnotes, so either the two numbers were written against different bases or the enforcement budget is scoped to a segment. The ratio survives either reading. The clamp-down allowance is about 0.84 percent of the projected scam and banned-goods revenue [4], one dollar forgone for every 119 booked [5].
The 95 percent bar is the part with an attacker on the other side of it. A ban gate set at that level of machine certainty is tuned to produce false negatives, and Reuters describes what happens in the gap: accounts Meta believes are likely scammers but cannot certify keep running and pay a higher rate [3][4]. Reuters reports the higher rate is intended to discourage those buyers [5]. Tom Uren, writing in Seriously Risky Business, reads the incentive as two-sided, because the scammer's penalty arrives as Meta's revenue [9]. For a fraud operator the working effect is that detection short of certainty converts into cost of goods sold instead of account loss.
Brand impersonation is where this reaches most security teams. Ads that Meta's legal team scored as carrying "higher legal risk", including impersonation of a brand or celebrity, produced $3.5 billion every six months [6]. That annualises to about $7 billion, roughly 44 percent of the $16 billion projection [6][7]. Anyone running a brand abuse takedown program is filing into that pipeline. A report carrying enough evidence to clear the confidence threshold removes an advertiser; one that falls short leaves the advertiser live at a higher rate.
The documents also show management weighing the windfall against the cost of regulatory action [12], with one noting the revenue would almost certainly exceed the cost of any regulatory settlement involving scam ads [7]. That line is the one to model against. A fine sized to a settlement has already been priced. Rob Leathern, a former Meta employee, told Wired that platforms should be made to relinquish money earned from scam ads, with the proceeds funding anti-scam non-profits [11]. It is the only remedy in the source material that changes the arithmetic above.
There is nothing to patch here, and the most quoted number is the softest: Uren flags the ambiguity in the safety staff estimate that Meta platforms were involved in a third of successful US scams [2], since much of that involvement may be WhatsApp carrying conversations with victims rather than ads supplying the lure [10]. The advertising figures carry less ambiguity, because Meta priced them itself.
Ranked by verification strength, evidence, and original report placement.
Reuters, working from a cache of documents it reviewed, reported that Meta had projected its 2024 advertisements for scams and banned goods would bring in about USD$16 billion, or 10% of its total revenue.
In one of the documents, Meta safety staff estimated that the company's platforms were "involved" in a third of all successful scams in the US.
Other documents revealed that Meta only bans advertisers if its automated systems are 95% certain that an account is committing fraud.
If an account does not meet the 95% threshold but Meta still believes it is likely a scammer, the company charges higher advertisement rates as a "penalty".
According to Reuters, the intent of the higher advertisement rates is to discourage suspicious advertisers from buying ads.
Meta took in $3.5 billion every six months from ads that its legal team had determined carried "higher legal risk", such as impersonating a brand or celebrity.
Distinct publishers with included, body-backed reporting in this cluster.
1 article · September 2, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
product
Reuters found 39 more banking-malware ads still running on Meta after India's advisory1 distinct publisher
product
Meta scraps the AI usage test from its performance reviews1 distinct publisher
security
Washington names industrial-scale distillation, then hands the detection bill to abuse teams1 distinct publisher
invest
Meta's Hatch agent tops out at $199.99 a month, with DoorDash and Etsy behind the meter2 distinct publishers
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Specific figures, second-hand chain
Every number here — the $16 billion projection, the 95% ban threshold, the $3.5 billion half-year, the $135 million clamp-down budget — originates in documents only Reuters has seen, and reaches us one step further removed through Seriously Risky Business. Nothing is quoted at length, no document is dated beyond the February 2025 plan, and Meta answers nowhere in this coverage. The internal consistency check also wobbles: the two percentage anchors imply revenue bases about $70 billion apart.
Nothing here to count
There is no product, patch or announced policy change whose uptake could be tracked. This reporting does not say whether the 95% bar moved, whether the $135 million was ever spent, or whether any regulator acted on the documents — so we leave this unscored rather than dress silence up as a signal.
Big number, cautious narrator
Unusually, Seriously Risky Business argues down its own headline, asking how much of the "third of all successful scams" claim is just WhatsApp carrying conversations rather than ads doing the baiting. What still tips slightly overstated is the $16 billion: it is an internal projection of ad spend attributed to scams and banned goods, not booked and audited revenue, and the accompanying percentages do not reconcile. The commentary is more careful than the arithmetic it repeats.
The penalty is the margin
The structural point is the sharpest thing in this reporting: charge a suspected scammer more and their deterrence fee becomes Meta's revenue. The documents make the same calculation explicitly, comparing ad income with the price of any regulatory settlement, and the forbearance budget lands at roughly 119 to 1 against the projected take. Worth naming the venue's own position too — a sponsored weekly newsletter that says outright it holds a cynical view of platforms and endorses forced disgorgement.
Trust the mechanics, not the totals
The mechanics are the durable part: a 95% certainty bar and penalty pricing beneath it are hard details to garble in retelling, and there is no sign the newsletter misreads Reuters. Confidence stops climbing there — one publisher, one investigation, no company response, no independent look at the cache, and dollar figures whose meaning shifts depending on whether they describe a projection, gross ad spend, or revenue actually booked.