Between June 2024 and September 2025, the Visa and Mastercard ATM fee class action settlement received 63.5 million claims—but 63.2 million were flagged as fraudulent and recommended for rejection, leaving only 296,877 valid claims approved for payment. This extraordinary 99.5% fraud rejection rate occurred because the settlement administrator deployed advanced fraud detection technology including ClaimScore analysis and internal data cross-checks that identified duplicate submissions, fabricated transaction histories, automated bot submissions, and claims from ineligible cardholders. The combination of loose initial eligibility requirements and sophisticated bad-faith actors attempting to exploit a $197.5 million settlement created a perfect storm where nearly every claim faced scrutiny.
This article explains why the fraud rate soared to unprecedented levels, how claims were actually verified, what makes a claim eligible for payment, and when the small percentage of approved claimants can expect to receive their share of the settlement. The scale of rejection is almost impossible to comprehend: if the fraud findings are upheld by the court, only about 0.47% of claimants will receive payment. This isn’t a case where the settlement administrator was overly aggressive or made honest mistakes—the volume of demonstrably fraudulent submissions was staggering, and independent analysis confirms the fraud detection methodology was sound. Understanding why your claim may have been rejected, and whether you have grounds to challenge the decision, requires knowing both how the system worked and what actually qualifies as a valid ATM fee overcharge.
Table of Contents
- Why Did 99.5% of ATM Settlement Claims Get Flagged as Fraudulent?
- How the Settlement Administrator Actually Identified Fraudulent Claims
- What Made Millions of Claims Ineligible Before Fraud Detection Even Began
- The Role of ClaimScore Technology and Automated Decision-Making
- Denial Reasons and the Reality of the Appeals Process
- Settlement Payout Details and the Approved Claimants’ Timeline
- The Future of Settlement Fraud Detection and Lessons Learned
Why Did 99.5% of ATM Settlement Claims Get Flagged as Fraudulent?
The atm fee settlement was designed to compensate customers who paid surcharges when using out-of-network ATMs between 1995 and 2024. The initial claim process required minimal documentation—claimants simply submitted their ATM transaction details, and the settlement portal accepted claims with relatively light scrutiny at intake. this created an irresistible target for fraudsters. Within months, the claim volume exceeded what the defendant banks could plausibly have charged in fees during the entire lawsuit period, and massive numbers of claims contained obvious red flags: identical transaction amounts submitted by different claimants, claims from the same household appearing dozens of times, transactions dated outside the settlement period, and claims from cardholders whose banks’ records showed zero ATM fees charged. The settlement administrator implemented ClaimScore technology and extensive internal data verification to catch the fraud.
These systems cross-referenced submitted claims against actual banking records from Visa and Mastercard, identified duplicate claim IPs and device signatures, and flagged patterns consistent with automated submission tools. What emerged was evidence of coordinated fraud campaigns—some claims were clearly submitted by bots, others by individuals filing hundreds of claims under stolen identities or family members’ details. The fraud detection found that approximately 40% of submitted claims contained at least one major red flag indicating they were either completely fabricated or that the claimant had already been reimbursed by their bank for the disputed fee. This wasn’t fraud of the magnitude that occasionally slips through—the scale and obviousness suggest that a significant portion of claims came from spam operations, scam websites promising “easy ATM settlement money,” and coordinated rings that harvested or purchased lists of card numbers. Unlike typical settlement fraud where bad actors file maybe a dozen claims, this settlement saw individual submitter profiles with hundreds or thousands of submissions, making mass filtering a necessity rather than an overcorrection.
How the Settlement Administrator Actually Identified Fraudulent Claims
The fraud detection process combined three overlapping verification methods. First, the administrator cross-checked every submitted claim against Visa and Mastercard’s internal records from the settlement period, looking for actual ATM transactions and surcharge records. If a claimant submitted a transaction that didn’t exist in the card network’s database, or if the claim described a fee the cardholder’s bank had already refunded, the claim was immediately flagged. This catch-all eliminated most fabricated claims and duplicate reimbursement attempts. However, if you genuinely used an out-of-network ATM and paid a fee that your bank never refunded, your claim would pass this verification—but only if the fee was actually charged and only if it occurred during the eligible period (1995-2024). Second, the administrator used pattern analysis and device tracking to identify coordinated fraud rings. Claims submitted from the same IP address, device, or email domain that appeared suspicious were grouped together and cross-referenced for duplicates. If 50 claims arrived from the same device submitting identical transaction amounts, the administrator could reasonably conclude these were fraudulent.
Similarly, claims that arrived in suspicious batches or used automated submission signatures (consistent keystroke patterns, form-filling bots, VPN masking) were flagged. This method was highly effective at catching industrial-scale fraud but occasionally caught legitimate claimants using shared networks—though those individuals could theoretically appeal with documentation. Third, the system flagged claims from people who submitted thousands of claims using slight variations on the same names or account details, a hallmark of professional fraud operations. One analysis found submitter profiles with 5,000+ claims using variations like “John Smith,” “Jon Smith,” “J. Smith,” all with slightly different addresses but the same core identity. The fraud detector’s job was to identify that pattern and mark all associated claims for rejection. The limitation here is that a claimant who legitimately filed multiple times accidentally, or who resubmitted after initially being rejected, could sometimes be caught in this automated filtering. However, this was supposed to be addressed through the appeal process.
What Made Millions of Claims Ineligible Before Fraud Detection Even Began
Beyond the fraud filters, millions of claims failed basic eligibility checks that have nothing to do with authenticity—they simply didn’t meet the settlement’s requirements. The settlement explicitly covers only debit card ATM surcharges from transactions between 1995 and 2024 in the United states. This immediately disqualified any claims involving credit cards, cash advances, prepaid cards, or ATMs outside the U.S. Many claimants submitted fees for checking account overdraft charges (which aren’t the same as ATM surcharges) or convenience fees for non-ATM services, which were also ineligible. Another major eligibility issue: if a claimant’s bank had already refunded the disputed ATM fee as a goodwill gesture or through the bank’s own dispute resolution process, the settlement explicitly excludes those claims. Visa and Mastercard aren’t liable for fees their customers’ banks already made customers whole on.
The settlement administrator checked this by cross-referencing banks’ refund records, and millions of claims failed this test. For example, if you paid a $3 ATM fee in 2019, complained to your bank, and received the $3 back in 2020, your claim to the settlement is ineligible because you’ve already been compensated. Additionally, claims from customers whose banks are registered outside the United States were ineligible, which ruled out claims from many expatriates or people who maintained foreign accounts. The rationale was that the settlement only applied to U.S.-based card issuers, and enforcing eligibility based on where the bank is chartered was simpler than trying to verify international banking records. This created a category of people with legitimate U.S. ATM surcharge claims who were nonetheless excluded purely due to how their bank was structured.
The Role of ClaimScore Technology and Automated Decision-Making
The settlement administrator’s use of ClaimScore and automated fraud detection created efficiency at the cost of some precision. ClaimScore is a proprietary system that assigns risk scores to claims based on dozens of variables: transaction patterns, temporal clustering, linguistic analysis of how claims were submitted, device signatures, and comparison to known fraud patterns from previous settlements. A claim flagged with a high ClaimScore doesn’t automatically mean it’s fraudulent in the criminal sense—it means the probability calculation weighted the evidence heavily toward fraud. However, the challenge for claimants is that explaining away a high fraud score requires access to documentation and an appeal process that many people don’t navigate successfully. One practical example: a retired person who decided to file a claim about ATM fees from 2000-2010, submitted all 30+ of their valid claims at once over a two-week period, might trigger the automated system’s pattern-detection algorithms.
Thirty claims in two weeks from the same submitter, with similar transaction profiles, could look like batch fraud rather than one person finally getting around to claiming legitimate fees they remembered paying. The automation doesn’t easily distinguish between a person who just learned about the settlement and proactively filed all their historical claims, versus a fraudster running a script. This is a real limitation: legitimate claimants who bunched their submissions together faced higher fraud scores than those who trickled submissions over months. The tradeoff was explicit: the settlement administrator could either accept the fraud rate (which was already 99.5%) and potentially pay out hundreds of millions in fraudulent claims, or enforce automated detection knowing that some small percentage of valid claims would get caught up. Given the volume—63.5 million claims in total—manual review of even a fraction of them was infeasible. The decision was to flag aggressively and allow appeals, rather than err on the side of trusting submissions.
Denial Reasons and the Reality of the Appeals Process
The most common denial reasons break into three categories: fraud flags (63.2 million claims), ineligibility due to claim type or transaction characteristics, and data verification failures. A claim can be denied for multiple overlapping reasons. For instance, a claim might be flagged for both fraud (duplicate submission from the same device) and ineligibility (credit card transaction rather than debit card). The settlement portal was supposed to provide detailed denial notices explaining which specific factors caused the rejection, but claimants reported that some denial letters were generic or lacked sufficient detail to mount a meaningful appeal. The appeals process exists, but it’s not straightforward. Claimants who received rejection notices can request manual review if they have documentation—bank statements showing the ATM fee, credit card transaction records, or contemporaneous emails proving the transaction occurred.
However, appeals require claimants to actively respond within a specified timeframe (typically 30-60 days), gather their own documentation, and submit a detailed explanation of why their claim should have been approved. Many claimants never learned that appeals were possible, or they didn’t have 20+ year old bank statements to support their claims. The warning here is that an initial rejection is not final, but the burden falls on the claimant to prove their case. A critical limitation: the settlement administrator’s burden of proof is minimal. They don’t have to prove your claim is fraudulent; they just have to show that automated algorithms identified risk factors. If you want to win an appeal, you generally have to prove your claim is legitimate—and if you paid the fee in 1997, you probably don’t have the documentation. This reversal of burden, while understandable given the fraud scale, means that technically valid claims sometimes go unpaid simply because the cardholder can’t retroactively prove what they paid 25+ years ago.
Settlement Payout Details and the Approved Claimants’ Timeline
The $197.5 million Visa and Mastercard settlement creates a compensation fund that’s divided among the 296,877 approved claimants. That averages to approximately $665 per approved claim—not a life-changing amount, but meaningful compensation for a service fee dispute. The actual payout per claimant will vary based on the number and amount of eligible fees they submitted; someone with 100 verified $2 fees will receive more than someone with one $3 fee. However, exact per-claim amounts haven’t been finalized pending court approval. Payment timing depends on court approval of the settlement agreement and the administrator’s fraud findings. Current projections from the settlement administrator and court documents indicate that the first wave of payments to approved claimants is expected in late winter 2026 (February-March 2026).
This assumes the court approves the fraud findings and settlement distribution plan without major delays or legal challenges. If the class appeals the fraud detection methodology or if the court requires additional review, payments could slip into spring 2026. Importantly, rejected claimants won’t receive any distribution unless they win a subsequent appeal—the 296,877 approved claims represent the only group in line for payment under the current settlement. a separate $167.5 million Visa and Mastercard settlement is addressing surcharges at independent, nonbank ATMs (as opposed to bank ATMs, which this settlement covers). That second settlement had preliminary approval in late 2025 and may distribute payments independently of this settlement, though timelines haven’t been finalized. If you were charged surcharges at nonbank ATMs and received a rejection from this settlement, you may be eligible for the separate independent ATM settlement.
The Future of Settlement Fraud Detection and Lessons Learned
The ATM settlement’s 99.5% fraud rate has become a cautionary tale in the settlement administration industry. Future large-dollar settlements will almost certainly implement fraud detection from day one, rather than accepting submissions loosely and filtering fraudulent claims afterward. The fraud volume was so extreme that it has prompted class action attorneys and settlement administrators to collaborate on pre-submission identity verification, requiring claimants to verify their identity or provide documentation upfront, rather than during appeal.
Additionally, cryptocurrency ATM fraud (a separate phenomenon involving ATMs installed by scammers that steal funds) caused $333 million in losses in 2025 alone, and this has heightened overall scrutiny of ATM-related fraud schemes. Regulators and settlement administrators are increasingly aware that the ATM space attracts sophisticated fraud operations, so any future ATM-related settlements or class actions will likely require stronger identity verification and documentation at claim intake. For claimants in this settlement, the lesson is clear: if you received a rejection notice and believe it was in error, you have a window to appeal with documentation. Waiting and hoping the decision will be reversed on its own won’t work—the appeals process requires active participation.
You Might Also Like
- How Much Will ATM Settlement Pay Per Person After Fraud Claims Removed
- ATM Settlement Still Processing Claims What Happens Before Payments Go Out
- Why You Haven’t Received Your ATM Settlement Money Yet Full Breakdown
Open Settlements You Can Claim Now
Browse current class action settlements accepting claims — several require no proof of purchase: