Input data and information sources
The person initiating the payment determines its amount, recipient, device and location; that person may be a fraudster. Each payment is linked to a card or account, and therefore to a person.
Payment fraud detection
A model assesses each payment for suspected fraud in real time and stops suspicious payments or refers them for review.
Understand the use case
What defines this use case?
Payment networks, payment service providers and banks use these models. The European Banking Authority and BaFin identify this as one of the most common AI uses in finance.
This page describes a typical system design. Actual data flows and permissions depend on your implementation.
- System design
- Analysis and processing
- Ability to act
- Can act within applications
- Fields of use
- banking, card payments, payment services
- Also known as
- fraud detection, transaction monitoring
How it works, step by step
System design and data flows
Input data
Each payment supplies its amount, recipient, time, device and location, alongside the card or account's payment history.
The person initiating the payment determines its amount, recipient, device and location; that person may be a fraudster. Each payment is linked to a card or account, and therefore to a person.
Processing
A model compares the payment with known fraud patterns and normal behaviour and calculates a risk score.
The system compares transactions and their characteristics with learned patterns or expected behaviour. It produces an anomaly assessment that can trigger further review. Blocking or declining a transaction occurs only through subsequent process logic and the authority assigned within it.
Results
The system approves the payment, requests additional confirmation, stops it or rejects it.
The result describes the defined model task. A match, category or forecast acquires its professional significance only in conjunction with the underlying data and intended use.
Use and impact
The decision is made while the person waits at a checkout or online shop and determines whether the payment goes through.
Write or execution permissions allow the system to change a state within an application. Model responses, tool calls and changes actually executed must therefore be considered separately.
Explore the system design: Select a component to see its task, incoming and outgoing data, and role in the use case. The guided walkthrough explains each phase. Then use Yes, Uncertain and No to represent your specific system variant.
12 review first6 clarify7 baseline threats
- Present
- Depends on your system
- Absent
- Review first
- To clarify
How to use the model
The diagram shows the typical architecture. Select a component to see what happens there and adapt its features to your system. Enable “Analysis guidance” to see which threats need reviewing for this architecture. The page's checklist continues to describe the typical case.
Changes from the typical case
None yet. Each change will appear here.
Where most threats arise
- 4 review first
- 3 review first, 1 clarify
- 3 review first
- 1 review first, 1 clarify
In this system design: 12 to review first, 6 to clarify, 7 baseline threats.
Architecture as a list
External access: Present
Review first
To clarify
Input: Part of every AI system
Review first
Third-party content: Present
AI model: Part of every AI system
Review first
- Model Drift & Degradation Review priority
To clarify
Provider service: Depends on your system
To clarify
Document collection: Depends on your system
To clarify
Training and feedback: Present
Review first
Result: Part of every AI system
Baseline for every AI system
Business application or equipment: Present
Review first
- Excessive Agency Review priority
person: Part of every AI system
Baseline for every AI system
Affected person: Present
Unapproved AI services: Depends on your system
To clarify
Supply chain and technical foundations: 5 baseline threats applying to every AI system
Baseline for every AI system
How it differs from similar use cases
Money laundering detection and sanctions screening
Searches across many entries for money laundering and checks parties against sanctions lists
Credit scoring and creditworthiness assessment
Assesses a person's ability to repay rather than individual payments
AI-based cyberattack detection
Detects compromised accounts and phishing in networks and emails
Data and information
Processing and operating environment
When a bank or merchant uses a payment service provider, the model comes from that provider.
Storage and further use
Payment networks and service providers train their models using payments that pass through them. Chargebacks and manual review results feed back as new training examples.
Results and data sharing
The system stops or rejects high-risk payments before a person sees them.
Using a provider does not automatically determine where processing takes place. Storage location, subcontractors, deletion periods and use for training must be checked for the specific deployment.
Assess risks and determine measures
From understanding the system to analysing risks
- Define the system boundary
Record data, participants, interfaces and permissions. Resolve uncertain characteristics in the model.
- Investigate threats
Relate review candidates to actual data flows: what could happen, and who would be affected?
- Assess risk
Assess impacts and likelihood in the deployment context. The number of review candidates is not a risk score.
- Provide evidence of measures
Select appropriate controls, assign responsibilities and define effectiveness checks. Document residual risks and approvals.
Key threats in the application context
- 4 particularly important
- 8 typical
- 6 depending on the system
- 7 additional threats for every AI system
A total of 25 threats from the threat catalogue may be relevant to this use case.
Also review: whether the system disadvantages groups
Check whether selection, ranking or assessment disadvantages people with protected characteristics, such as sex, age or origin.
Priority review points
Fraudsters use small payments to test what gets through, then adjust the amount, merchant and timing. The model subsequently treats later payments as unremarkable.
New scams, such as transfers following a call from a fake bank employee, differ from learned cases. Detection is poorer until enough new examples are available.
Targeted Poisoning / Label Poisoning
Customers who report their own purchases as fraud introduce incorrect labels into training data. The model then flags similar legitimate payments or misses genuine fraud.
The system can reject payments without human review. A wrongly configured threshold can stop many legitimate payments in a short time.
8 further review candidates in the typical design
Other relevant threats
Your own training or operational feedback can introduce manipulated data into training datasets.
Listed because of these characteristics
- Learns from your own data
- Uses operational feedback
Training data or an adopted base model can introduce hidden triggers for misbehaviour.
Listed because of this characteristic
- Learns from your own data
Anyone able to query the model can replicate it through many queries. Internally trained weights can be stolen through network access.
Listed because of these characteristics
- Learns from your own data
- Accessible over a network
Complex or high-volume inputs can exhaust model computing resources.
Listed because of this characteristic
- Accessible over a network
Targeted queries can recover some personal or protected training data from the model.
Listed because of these characteristics
- Processes personal data
- Processes confidential data
Sensitive Information Disclosure
Confidential content can appear in answers, supplied context or logs.
Listed because of these characteristics
- Processes personal data
- Processes confidential data
Exfiltration from ML Application
Protected data can leak through tools, network connections, storage or insiders.
Listed because of this characteristic
- Processes confidential data
A network-accessible application can be overwhelmed by mass requests.
Listed because of this characteristic
- Accessible over a network
Up to 6 further threats may apply, depending on your system design. Nine threats need to be reviewed for every AI system. Two of these are particularly important here and are therefore listed above.
System features and open questions
Each Yes adds further threats to your checklist, up to 6 in total. If you cannot yet answer a question, include its threats in your review.
- Added by this Yes
How your system is built
- Uses a knowledge graph1 threat
- Optimises a specified metric1 threat
How you operate the system
How your system is built
Does the system use a knowledge graph storing concepts and their relationships?
Some systems represent accounts, devices and merchants as a network of relationships to detect connected fraud rings.
If yes, add
Is the system intended to maximise a specified metric, such as revenue, utilisation or clicks?
If the deployer sets the threshold according to fraud costs and lost revenue, the system pursues a business objective.
If yes, add
How you operate the system
These questions arise for almost every AI system.
Can people or programs query the model directly, for example through an interface?
If yes, add
Do you pay per query or unit of computing time, so costs increase with use?
If yes, add
Does the system share servers or computing resources with other customers or applications?
If yes, add
Can people outside your organisation use the system, including without login?
If yes, add
Do staff also use AI services for this task that your organisation has not approved?
If yes, add
Controls and implementation
Measures for the priority review areas are expanded. Each linked measure explains how to implement it.
Review priorityAdversarial Inputs2 measures
Review priorityModel Drift & Degradation3 measures
Review priorityTargeted Poisoning / Label Poisoning3 measures
Review priorityExcessive Agency3 measures
Training Data Poisoning2 measures
Backdoor ML Model2 measures
Model Theft3 measures
Model Denial of Service3 measures
Privacy Attacks3 measures
Sensitive Information Disclosure3 measures
Exfiltration from ML Application3 measures
Application Denial of Service3 measures
Controls for the remaining threats are available on their pages in the threat catalogue.
AI Act provisions
Applies to every AI system
Art. 4: AI literacy
Applies if staff or appointed representatives operate or use AI systems.
Applies since 2 February 2025.
Obligations and details for Art. 4: AI literacy
- Who bears the obligations
- Provider, Deployer, Product manufacturer, Provider of a general-purpose AI model
Detailed obligations
Generally does not apply
Annex III no. 5 point b: Creditworthiness
Annex III, point 5(b), expressly excludes systems intended to detect financial fraud.
Obligations and details for Annex III no. 5 point b: Creditworthiness
- The general condition is
- Applies if the system assesses people's creditworthiness or establishes their credit scores.
- Exception
- Systems intended to detect financial fraud.
- Who bears the obligations
- Provider, Deployer, Product manufacturer
Detailed obligations
Classify your system with the quick checker All obligations in the obligations compass
Cross-cutting threats
Review these threats for every AI system, whatever its purpose.
Atypical system features
View 15 further questions
If any of these features does apply to your system, add the listed threats.
Does the system generate its own text, such as answers, drafts or summaries?
The system produces a risk score and a decision, rather than text.
If yes, add
Does the system generate or edit images, sounds, voices or videos?
If yes, add
Does the system create new content from open-ended requests that people read, see or hear directly? Translation, reading aloud and transcription do not count.
If yes, add
Can users give the system natural language tasks that it then performs?
The model assesses payment data and does not accept natural language instructions.
If yes, add
Does the system search its own collection of documents or records for relevant content?
If yes, add
Does the system formulate answers from retrieved documents?
If yes, add
Does the system remember content from earlier conversations or tasks and reuse it later?
If yes, add
Does the system call other programs, interfaces or search services itself, even if only to read?
If yes, add
Does the system determine its next steps itself based on earlier results?
If yes, add
Does the system delegate tasks to other AI agents or incorporate their results?
If yes, add
Is the system connected to tools or data sources through the Model Context Protocol (MCP)?
If yes, add
Do documents from several customers, departments or permission groups share one search store?
If yes, add
Does the system process several input types together, such as text with images or speech?
If yes, add
Does the system generate code, scripts or database queries for further use?
If yes, add
Does the system execute programs, scripts or commands itself?
If yes, add
Sources
- Report of a deployment
Visa, Visa Prevents Approximately $25 Billion in Fraud Using Artificial Intelligence, 17 June 2019 According to the deployer
Visa describes using AI to assess all VisaNet authorisations in real time.
- Report of a deployment
Stripe, Radar: Risk evaluation (documentation) According to the deployer
Stripe describes machine learning models that assess each payment's risk in real time.
- Authority, court or parliament
EBA, Follow-up report on machine learning for IRB models, 2023
The European Banking Authority identifies fraud detection as a major use of machine learning in banking.
- Authority, court or parliament
BaFin lists fraud detection among the most common AI uses in finance.
Related use cases
Also in the field of Banking and insurance
- Credit scoring and creditworthiness assessmentA model estimates repayment likelihood from payment and customer data, helping determine whether a person receives credit and on what terms.
- Money laundering detection and sanctions screeningBanks search account and payment data for money laundering patterns and check all parties against sanctions lists.
- Life and health insurance risk assessment and pricingInsurers estimate an applicant's mortality or illness risk to determine whether to accept them and what premium to charge.
- AI-assisted insurance claims handlingThe system analyses claim documents and photographs, checks entitlement against the policy, flags inconsistencies and prepares settlement.
Often used together with
- Biometric verification and video identificationThe system checks whether a person is who they claim to be, for example by comparing a live image with their ID photograph.
- Risk-based review selection for applications and benefitsAuthorities assess applications or payments for error or fraud risk and use this to select cases for closer review.
Updated: 30 September 2026. Review checklist rules version 0.3.2, threat catalogue version 2026.10.24.2. How these pages are structured