Theme

Fraud Detection & AML

Transaction monitoring, identity verification, sanctions screening

Filter in Explore

Cited evidence

Reported outcomes in this theme

Modeled result

Mastercard

Decision Intelligence Pro fraud scoring

+20% average fraud-detection uplift in initial modeling

Company-reported measure

HSBC

Dynamic Risk Assessment sharpens AML review

60% fewer false-positive cases reported by HSBC

Company estimate

Mastercard

SafetyNet provides a network fraud backstop

$77bn company-estimated fraud losses prevented cumulatively since 2020

104 researched workflows · provenance on each case

109 catalog records

Universal Credit Advances fraud-review prioritisationDWPThe live model risk-scores benefit advance requests and refers high-risk cases for human assessment; it does not decide entitlement.Scaled · reported stage
Transaction fraud screening and merchant monitoringNetwork InternationalNetwork International uses Brighterion's learning models for merchant-acquiring fraud screening and monitoring across its portfolio.Scaled · reported stage
Stripe Radar ML Fraud DetectionStripeStripe Radar uses AI trained across its business network for real-time payment-fraud detection and blocking. Stripe describes Radar as available within its payment processing and through signals that merchants can use in their fraud workflows.Scaled · reported stage
AML client-conversation summariesABN AMROVoice AI summarizes client discussions for AML analysts.Production · reported stage
Corporate KYC entity resolution and network analysisABN AMRO Bank N.V.ABN AMRO's KYC analysts use entity resolution and graph analytics to connect corporate ownership and relationship data, supporting complex client investigations.Production · reported stage
Contextual website screening during KYC onboardingAirwallexAirwallex uses LLMs to interpret business websites contextually during onboarding, reducing ambiguous keyword alerts in due diligence.Production · reported stage
Incognito suspicious-claim referralAllianzMachine learning flags potentially fraudulent UK claims for specialist investigation.Production · reported stage
Amazon multimodel detection of fake reviews and coordinated account abuseAmazonAmazon combines machine learning, language models and graph neural networks to detect review manipulation. Suspicious cases lacking sufficient evidence go to trained investigators.Production · reported stage
Global real-time card transaction fraud modelsAmerican ExpressNVIDIA describes American Express monitoring global customer transactions with fraud algorithms and improving model accuracy.Production · reported stage
AI-assisted entity and relationship extraction from large document corporaATOANAO records ATO's January 2025 use of commercial software incorporating AI to extract intelligence from large structured and unstructured datasets; predecessor models had been decommissioned.Production · reported stage
Aviva AI-assisted suspicious-claims detectionAvivaAnalytics and AI tools help identify suspicious claims and manipulated evidence for human review.Production · reported stage
Digital banking fraud-pattern detectionBanco BPIUses Digital Trust to identify suspicious behavioral patterns and enhance real-time fraud prevention and customer authentication.Production · reported stage
Unified behavioral fraud monitoringBanco BVImplemented behavioral biometrics, transaction monitoring and threat intelligence across onboarding and payments to detect account takeover and suspicious device activity.Production · reported stage
AI fraud prevention and detection modelsBank of AmericaBank of America says it has implemented more than 50 AI models to help prevent and detect fraud. The issuer confirms operational use without quantifying transaction coverage, real-time latency or avoided fraud losses.Production · reported stage
Machine-learning card-fraud decisions supported by AerospikeBarclaysBarclays consolidated fraud rules, models and customer profiles into a shared machine-learning fraud platform supported by Aerospike. The vendor describes operational card-authorization decisions and lower latency.Production · reported stage
AI identity-document verificationBOCHKExternal AI checks identity-document images for tampering during mobile account opening.Production · reported stage
Pix and card fraud monitoring with biometricsBTG PactualIntegrated Feedzai fraud controls, device recognition and facial biometrics as it expanded into full banking and card/Pix transaction services.Production · reported stage
Graph machine learning for authorized push-payment fraudCapitecCapitec creates transaction-network features in Memgraph and scores records through SageMaker, supporting detection of connected authorized push-payment fraud patterns.Production · reported stage
Alice procurement-document risk alertsCGUCGU uses text mining and AI to examine procurement notices daily and alert staff to potential procurement risks or inconsistencies.Production · reported stage
Schwab machine-learning fraud detectionCharles SchwabSchwab reports longstanding use of machine-learning models for fraud detection across its business.Production · reported stage
Chime fraud detection and risk scoringChimeChime uses AI fraud and risk models, retaining rules and human oversight for enforcement.Production · reported stage
Identity-first credit application fraud decisioningCNG HoldingsCNG passes credit applications through a SAS identity platform using analytics and machine learning, joining external data and evaluating whether the applicant owns the claimed identity.Production · reported stage
Bounded LLM review of Onramp transaction sequencesCoinbaseCoinbase deployed selective LLM review after existing fraud models/rules. The agent can add a block but cannot overturn an upstream block; application code enforces decisions.Production · reported stage
ML-assisted fraud-rule recommendation and backtestingCoinbaseRisk analysts use ML/statistical recommendations to select rule parameters and test them against historical transactions before deploying fraud controls.Production · reported stage
Identify investment-scam wallets with AlteryaCoinbaseCoinbase's payment and operational risk leader describes using Alterya to identify wallet addresses linked to investment scams and protect customers.Production · reported stage
Fraud-pattern and rule-generation agentCommBankAgent detects emerging payment patterns and proposes detection rules for fraud-analyst approval.Production · reported stage
Apate scam-intelligence botsCommBankConversational bots engage scammers and supply intelligence into the bank's scam controls.Production · reported stage
Card-program fraud decisioningCoreCardUses Feedzai machine learning for fraud detection and client-configurable rules and review queues.Production · reported stage
Credit-applicant network fraud analysisDBS BankNetwork-graph analysis helps identify potentially fraudulent credit applicants.Production · reported stage
Discovery Life suspicious-claim triage and assessor supportDiscovery LifeIn-house ML and graph analytics flag unusual claims for human investigation; GenAI structures documents for assessors.Production · reported stage
Real-time self-learning fraud preventionDNBAI decisioning and self-learning support transaction fraud detection.Production · reported stage
Issuer-specific machine-learning fraud modelsEloMigrated issuing clients to a multi-tenant fraud platform and worked with Feedzai to build machine-learning models for each issuer.Production · reported stage
Expedia real-time malicious-activity detectionExpedia GroupExpedia reports AI systems analyzing data to identify and prevent malicious activity affecting travelers and partners.Production · reported stage
Machine-learning identification of potential scam websitesFCAFCA uses machine learning to screen newly registered web domains for potential financial scams.Production · reported stage
AI transaction fraud scoringFNB / FirstRandA FirstRand data-analytics executive describes models assessing simultaneous transactions for fraud and returning decisions in milliseconds.Production · reported stage
Agentic EDD Analyst investigation preparationFNBOAI gathers due-diligence evidence and creates case documentation for human investigators.Production · reported stage
Merchant transaction screening before network submissionGlobal PaymentsThe integrated processor service uses AI/ML to screen payments for card testing and other fraud before they reach card networks.Production · reported stage
HDFC ERGO suspected-claim fraud screeningHDFC ERGOAI analytics flag claims suspected to be fraudulent for review.Production · reported stage
HDFC Life pre-claim investigation checksHDFC LifeGenAI performs initial maker-checker investigation checks using face-match and risk information.Production · reported stage
AI prioritization of transaction-monitoring alertsHolviHolvi uses AI to prioritize transaction-monitoring alerts and explain flagged risk factors within its screening and ongoing-monitoring program.Production · reported stage
AML AI suspicious-activity detection with Google CloudHSBCHSBC deployed Google Cloud AML AI to identify suspicious activity, with demonstrated use in the UK and Hong Kong and expansion planned elsewhere. The report attributes two-to-fourfold greater suspicious-activity identification and 60% fewer alerts to the deployment.Production · reported stage
Entity Research Copilot financial-crime triageHTLFGenAI gathers entity and adverse-news research and documents findings for investigators.Production · reported stage
Behavioral biometrics and malware-based fraud detectionIbercaja BancoImplemented Digital Trust to analyze device interactions and session patterns, supplementing other controls and informing fraud rules.Production · reported stage
Corporate transaction and session anomaly monitoringINGBehavioural analytics and ML flag payment anomalies; specialists review alerts and contact clients.Production · reported stage
Global Sanctions Manager AI-assisted alert clearingJPMorgan ChaseJPMorgan uses supervised machine learning with prior-decision matching and Pega workflows to triage and clear sanctions-screening alerts and route remaining cases for review. JPMorgan speakers distinguish this sanctions and client screening workflow from subsequent AML transaction modeling.Production · reported stage
Real-time AI fraud monitoringKCBKCB reports AI-based monitoring that flags suspicious activity in real time.Production · reported stage
X-Sight fraud and AML modernizationKeyBankKeyBank is fully live with fraud and AML capabilities, incorporating Integrated Fraud Management, ActOne case management and customer due diligence on the SaaS platform.Production · reported stage
Real-time fraud investigation agents for frontline teamsLloyds Banking GroupMultiple AI agents support fraud colleagues inside existing tools; colleagues retain final decisions.Production · reported stage
Network analysis to flag authorized push-payment scamsMastercardThe live UK service analyzes payer/payee histories and links to scam-linked accounts before transfers. TSB is explicitly described as using it.Production · reported stage
Predict compromised card details before fraudulent useMastercardMastercard combines transaction patterns and generative prediction to identify compromised cards from incomplete stolen-card information, enabling earlier bank intervention.Production · reported stage
Machine-learning e-commerce authenticationMetro BankMetro Bank uses TSYS Smart Secure with Featurespace behavioral machine learning to assess online purchase risk and support 3DS authentication decisions.Production · reported stage
Monzo adaptive fraud detectionMonzoMonzo describes live ML fraud defenses using multi-task deep learning for unauthorized transactions, scams and money-muling.Production · reported stage
Global AI-assisted customer screening and ongoing monitoringNissan Financial ServicesNissan Financial Services replaced fragmented market-level tools with unified screening and monitoring, using AI matching, refreshed risk data and automatic audit records.Production · reported stage
Insurance fraud-risk prioritisationNN GroupNN’s Fraud Risk Model identifies levels of potential fraud risk for employees to investigate.Production · reported stage
Unified digital fraud and AML monitoringNovobancoUses digital fraud monitoring; a 2025 agreement expanded transaction fraud capabilities and integrated AML into a unified fraud, KYC and AML platform.Production · reported stage
AML profile and network analysisOCBCAnalyses customer networks and transactions for suspicious communities.Production · reported stage
Adaptive ML screening of merchant card paymentsPayPalPayPal's available toolkit learns card-payment fraud patterns and accepts or rejects transactions; merchants can simulate and tune filters.Production · reported stage
Real-time merchant-payment fraud and risk detectionPaytmPaytm's annual report identifies Periscope as its AI fraud/risk platform.Production · reported stage
Merchant-specific payment fraud strategiesPayUCombines Feedzai detection with in-house expertise to customize fraud strategies by merchant and market, including Latin America.Production · reported stage
Piedmont cross-system drug-diversion detectionPiedmont HealthcarePiedmont uses machine learning across medication-management systems to flag discrepancies and high-risk behavior for investigators.Production · reported stage
AI behavior monitoring and AML case investigationsPleoPleo implemented AI-driven behavioral AML monitoring with case management and SAR workflows. Analysts use consolidated customer data to investigate suspicious activity.Production · reported stage
Cash-flow discrepancy monitoringPostFinanceAI monitors payment flows for suspicious discrepancies and money-laundering signals.Production · reported stage
Custom ML screening of gift-card transactions to prevent chargeback fraudRaiseRaise uses Sardine custom ML models together with tuned rules to detect gift-card fraud. The vendor/customer case reports chargebacks falling from 0.53% to 0.29% and manual reviews from 15–18% to 0.25% of transactions.Production · reported stage
AI-enabled government-ID verificationRBCAI supports digital identity verification for account opening.Production · reported stage
Risk models reduce unnecessary remittance interventionsRemitlyRemitly evaluates transaction patterns for identity theft and card fraud while minimizing verification friction for legitimate transfers.Production · reported stage
Revolut card-scam interceptionRevolutRevolut's ML feature identifies likely scam-related card payments, blocks risky activity and routes customers through intervention or specialist review.Production · reported stage
Santam real-time voice-impersonation detectionSantamAI analyzes voice interactions to identify fraud, with human oversight.Production · reported stage
AI automation of card-fraud claimsSantander BrazilSantander identifies Brazilian card-fraud claims as a deployed AI operations workflow.Production · reported stage
Project Protect trafficking-related AML intelligenceScotiabankML-supported profiling helps investigators find suspicious financial behavior associated with trafficking.Production · reported stage
Retail payment fraud detectionSociete GeneraleBehavioural AI applications identify external fraud and support blocking of suspicious payments.Production · reported stage
AI and ML name and transaction screeningStandard CharteredStandard Chartered integrates AI and machine-learning models into name and transaction screening to support timely, consistent financial-crime decisions and reduce manual interventions.Production · reported stage
Stripe Identity document and selfie verificationStripeStripe Identity uses machine learning to detect fake identification documents and spoofed photos and to match a live selfie to an identity-document photo. Stripe offers the production capability to businesses and uses the underlying technology in its own onboarding and risk processes.Production · reported stage
GAN-enhanced financial crime detectionSwedbankSemi-supervised deep-learning models improve fraud detection despite imbalanced transaction labels.Production · reported stage
Behavioral and device fraud analysisTBC BankImplemented AI-native Digital Trust to analyze behavior, devices and transactions during sessions and refine controls through feedback and testing.Production · reported stage
Tokio Marine Shift claims visual intelligenceTokio MarineA GenAI upgrade extracts and classifies image/document evidence in existing claims-intake and fraud-detection deployments.Production · reported stage
Treasury check-fraud detectionTreasury Fiscal ServiceMachine learning accelerates identification of fraudulent Treasury checks for payment recovery.Production · reported stage
Unified merchant transaction-risk operationsUnzerUnified risk monitoring across acquired businesses and uses Feedzai to refine models and rules. A separate TrustScore extension remains in pilot.Production · reported stage
Real-time digital fraud monitoringUOBMachine-learning systems monitor digital banking platforms for suspicious activity.Production · reported stage
Adaptive payment-fraud scoring across Verifone gatewaysVerifoneVerifone reports deploying adaptive machine learning to score payment risk across its gateways, modeling normal merchant, card and account behavior across channels.Production · reported stage
Zest Protect origination fraud screeningWauna FCUFraud models identify suspect applications and guide requests for supporting evidence.Production · reported stage
Real-time payment fraud and digital identity riskWio BankUses Feedzai AI and machine learning to protect instant payments and digital customer journeys.Production · reported stage
Real-time financial-crime risk alertsWiseWise trains ML models on customer profiles and transactions to generate real-time risk alerts for operating teams.Production · reported stage
Cloudera fraud-detection model pilotBRIData-science team combines datasets to develop real-time fraud monitoring.Pilot · reported stage
Fraud Assist call-investigation intelligenceFNBOGenAI summarizes call evidence and prepares fraud case documentation.Pilot · reported stage
Corporate false-invoice fraud detection pilotIntesa SanpaoloAI transaction models tested for tax-fraud and laundering patterns in corporate customers.Pilot · reported stage
KakaoBank joint federated-learning fraud modelKakaoBankFinancial Security Institute reports a tested joint voice-phishing detection model with KakaoBank, Toss Bank and Kbank, sharing model weights across institutions.Pilot · reported stage
Expert-reviewed initial fraud risk assessment draftingPublic Sector Fraud AuthorityFRA Accelerator retrieves curated fraud guidance and uploaded scheme context, uses a multi-step LLM workflow to draft and expand risk lines, then produces a structured spreadsheet for qualified assessors. The record describes PII redaction, per-session document isolation and deletion. It reports 143 users; capacity for up to 1,000 annual assessments is not completed assessment volume.Pilot · reported stage
Project Guardian synthetic-opioid AML modelingScotiabankBank experiments with behavioral models for fentanyl and synthetic-opioid financial investigations.Pilot · reported stage
Project Shadow AML transaction-pattern investigationsScotiabankML detects financial patterns potentially associated with child exploitation for investigator review.Pilot · reported stage
Real-time scam-call assistantWestpacAI supports call-centre operators handling potential scams.Pilot · reported stage
AI suspicious-activity and watchlist monitoringAberdeen GroupAberdeen selected ML/graph-based suspicious-activity monitoring and AI-assisted watchlist matching to strengthen financial crime investigations.Announced · reported stage
FOCAL financial-fraud preventionAl Rajhi BankAl Rajhi signed an agreement with Mozn to apply its AI financial-crime platform to fraud prevention and risk intelligence.Announced · reported stage
FIS AML Investigations Agent rolloutAmalgamated BankAmalgamated is deploying the Claude-powered agent to assemble AML evidence, contextualize alerts and assist investigation documentation. Analysts must approve closure and outputs link to underlying records.Announced · reported stage
Omnichannel fraud and AML platform selectionAmerican State BankThe Texas bank selected Xceed for onboarding, transaction monitoring, behavioral fraud detection and linked case investigations.Announced · reported stage
Ant Group AI Fraud PreventionAnt GroupAnt Group deploys AI-powered fraud prevention that processes billions of transactions on the Alipay platform. The system uses real-time machine learning to assess transaction risk, detect account takeovers, and prevent unauthorized payments with sub-second response times.Not assessed · legacy label
FIS Financial Crimes AI Agent developmentBank of MontrealBMO is named as an early development institution for the FIS/Anthropic agent assembling evidence and supporting AML investigator review.Announced · reported stage
Aurora collaborative financial-crime analyticsBISA completed first-phase experiment combined machine learning, network analysis and privacy technologies for collaborative anti-money-laundering analysis.Not assessed · reported stage
Hertha payment-system financial-crime analyticsBISBIS and Bank of England tested detection of coordinated illicit payment patterns using synthetic retail-payment data.Not assessed · reported stage
LLM-assisted off-chain risk intelligence researchEllipticElliptic announced ChatGPT integration for its own researchers and investigators to synthesize and organize intelligence about crypto risk factors.Announced · reported stage
Fannie Mae mortgage fraud detection partnershipFannie MaeFannie Mae announced a Palantir-powered unit to identify suspicious patterns in mortgage datasets and support investigations.Announced · reported stage
Transaction Monitoring AIGoldman SachsGoldman Sachs uses AI-powered transaction monitoring systems to detect suspicious financial activity across its global operations. The system applies machine learning to reduce false positive rates while maintaining high detection accuracy for potential fraud and money laundering.Not assessed · legacy label
AI-Driven AML/KYC ScreeningJPMorgan ChaseJPMorgan Chase deploys advanced AI models for anti-money laundering and know-your-customer screening processes. The system reduces false positives in transaction monitoring while improving detection of genuinely suspicious activity across millions of daily transactions.Not assessed · legacy label
Scam Check purchase-payment assessmentLloyds Banking GroupCustomer payment-journey AI assesses the likelihood of a purchase scam.Announced · reported stage
AI anti-fraud capabilities in M-PESA platform migrationM-PESA / SafaricomSafaricom announced transaction-pattern monitoring and SIM-swap quality screening as part of M-PESA's Fintech 2.0 upgrade.Announced · reported stage
Cross-bank scam-detection proof of valueMASMAS is testing AI/ML detection of risky accounts and transactions using historical data pooled from five banks.Not assessed · reported stage
AI Fraud Detection in ClaimsProgressiveProgressive employs AI models to detect fraudulent insurance claims by analyzing claim patterns, historical data, and behavioral indicators. The system flags suspicious claims for investigation, helping to reduce insurance fraud losses and keep premiums fair for honest customers.Not assessed · legacy label
AI identity verification for account openingSNBKnowledgeNet's signed contract covers analysis of customer stamps, signatures and documents to support identity verification and account opening.Announced · reported stage
Account-opening and payment anomaly detectionTruistAI models examine anomalies, threats and malware across onboarding and payment activity.Not assessed · reported stage
AI-Powered Financial Crime DetectionWells FargoWells Fargo uses AI to detect and prevent financial crimes including fraud, money laundering, and sanctions violations. The system processes millions of transactions in real-time, applying advanced pattern recognition to identify suspicious activity while reducing false positive alerts.Not assessed · legacy label