AI fraud detection- Service Overview | Developer

What is AI fraud detection?

AI fraud detection is a service that uses Payment information and machine learning to detect abuse. With more than hundreds of millions of Payment data records per year, it is now possible to create a model of all fraud patterns by machine learning and calculate real-time scores for similarities with fraud patterns that humans cannot distinguish. The model accumulates fraudulent transaction data, such as chargebacks, and relearns it.

Service Diagram

Available Payment Methods

Credit Card Payment

Main Features

Leverage machine learning
It uses machine learning on past fraud patterns to calculate the risk of fraud of the Payment as a score at the time the end user performs the Credit Card Payment. merchant enables early detection of fraudulent transactions by providing real-time scores.
Easy to implement
In many cases, the introduction of conventional fraud detection tools requires additional information from merchant, which causes the burden of introducing screens and revisions to terms and conditions, but "AI fraud detection" can reduce the burden of introducing merchant because it uses the information used by Payment.
Dynamic measures
It provides the ability to set unique rules merchant to prevent transactions suspected of fraud, and the ability to add EMV 3-D Secure authentication only to suspicious transactions.

Usage Information for End Users

For details on how to purchase Credit Card Payment, please refer to the service website.
For end-user usage information, please refer to the FAQ (Payment /System Specifications > Credit Card Payment).

Cautions

If you are using Link Type and want to use the rules that can prevent transactions that are suspected of fraudulent use, please Transition to the purchase error screen provided by our Payment or to the error specified URL specified by merchant.
If you use API type, you need to determine whether or not to provide the transaction after obtaining the detection result information such as the score returned by AI fraud detection on the merchant. If you determine that it is possible to provide it, please Payment it, and if you determine that it is not possible, please cancel Payment.

Service Specifications

Basic Specifications

The following Billing methods and system connection methods are available:

Billing Methods

One-time charge	○
Recurring charge (simple)	×
Recurring charge (based on term or usage rate)	×

* AI fraud detection is a service that detects unauthorized use when the end user enters Payment information or makes a transaction from our Payment screen or merchant website. merchant cannot be used with "Recurring billing (simplified)" or "continuous billing using the purchase API" that transacts at the time of payment.

system connection method

Link Type	○
API Type	〇

Supported devices

Please refer to the service overview of Credit Card Payment.
If you are using API type, please refer to the AI fraud detection IF specification.

List of Provided Functions

It offers the following features: The features of the AI fraud detection administration tool list only the main features. For details, please refer to the AI fraud detection Management Tool Manual.

Provided functions		Link Type	API Type
Scoring		○	○
Rule Judgment		〇	〇
gateway		〇	×^※¹
Additional Authentication (EMV 3-D Secure)		〇	×^※²
AI Fraud Detection Administrative tools	Dashboard	○	○
	Check the results	○	○
	Rule Settings	○	○
	simulation	○	○
	List Registration	○	○

*1: When using API type, it is necessary to obtain detection result information such as the score returned by AI fraud detection on merchant before deciding whether or not to provide the transaction.
*2: If you are using API type and would like to use additional authentication (EMV 3-D Secure), please contact our sales representative.

Details of each function are as follows:

Scoring
This function uses AI to calculate the risk of fraudulent use as a score in real time.

The score is calculated on a scale from 0 to 99, and the higher the score, the higher the risk the AI has determined the transaction to be.
You can view the score in the AI fraud detection administration tool.
If you are using API type, you can also get the detection results returned by AI fraud detection.
For more information, see the AI fraud detection IF specification.

Rules judge
This function allows merchant to set whether or not to provide a transaction after scoring and how to handle the transaction (e.g. obtain Authorization but set an alert to be raised).

You can select the type from 3-D Secure, Allow (Enforced), Block, Alert, and Allow.
Transactions that do not fall under any rule will be deemed "unjudged."
You can create the rules that are required to set up rule decisions in the AI fraud detection administration tool.
You can view the results of rule decisions in the AI fraud detection administration tool.
If you are using API type, you can also get the detection results returned by AI fraud detection.
If you use Link Type, the movement of our Payment screen will change depending on the result of the rule judgment.
For details, please refer to "Gateway".
When using API type, change the movement of merchant website according to the result of the rule judgment after obtaining the result of the rule judgment as the detection result returned by AI fraud detection.

Cautions

There are limitations to the number of rules that can be created. Please refer to "Restriction" in "Notes".
The creation of rules required for rule judgment settings and application of those rules must be carried out by merchant itself.
We do not make or enforce rules on your behalf.

"gateway"
This function changes the behavior of our Payment screen depending on the results of rule judgment.
If you are using Link Type, please check the movement of our Payment screen according to the results of the rule judgment below
If you are using API type, please change the movement of merchant website according to the result of the rule judgment.

Rule Judgment	Link Type Designation of "Payment Method"
Rule Judgment	Credit Card Payment	Credit Card Payment (EMV 3-D Secure)
3-D Secure	×^※¹	EMV 3-D Secure authentication
Allow (Enforce)	Authorization acquisition	Skip EMV 3-D Secure authentication and get Authorization^※3
block	Authorization acquisition discontinued^※²	Authorization acquisition discontinued^※²
Alerts	Authorization acquisition	Skip EMV 3-D Secure authentication and get Authorization^※3
permission	Authorization acquisition	Skip EMV 3-D Secure authentication and get Authorization^※3
No judgement	Authorization acquisition	EMV 3-D Secure authentication

*1: Although it is possible to create rules in the AI fraud detection management tool, the function to additionally authenticate EMV 3-D Secure only for transactions that are suspected of fraudulent use cannot be used. You need to apply for EMV 3-D Secure and switch to our Payment screen in Credit Card Payment (EMV 3-D Secure).
*2: You will be transitited to the purchase error screen on our Payment page or to the URL designated by merchant in case of an error.
*3: If you set the rules to "Allow (Forced)", "Alert", or "Allow", identity authentication (EMV 3-D Secure) will not be performed. Please note that if identity authentication is not performed and a chargeback (cancellation of Settlement proceeds) occurs due to a transaction made using Credit card information illegally obtained by a third party, merchant will be responsible for the cost.

<Additional Authentication (EMV 3-D Secure)>
This function allows you to add EMV 3-D Secure authentication by setting rule judgments.
By using this feature, you can enforce EMV 3-D Secure only on suspicious transactions.

《AI fraud detection Management Tools 》
It is a management tool that allows you to search and refer to the results of scores and rule judgments, and to create and simulate the rules necessary for setting up rule judgments. For details on how to operate the AI fraud detection management tool, refer to the AI fraud detection Administration Tool Manual.

Major Functions
Major functions are as follows:

Search/reference scores and rule judgment results
CSV output of displayed items
Various settings required for rule determination, such as "rule setting," "simulation," and "list registration"

Usage Overview

Relate to Admin portal / Payment management tool
Among the items that can be checked in Admin portal / Payment management tool, the items that can be checked in the AI fraud detection administration tool are as follows.

Tracking ID
Authorization results
Approval number
3D Authentication
Reason of Error
Order ID^※1
Customer ID^※1
Product ID^※1

*1: If you are using API type, the information you send with JavaScript to detect fraud will be displayed in the AI fraud detection management tool.

Payment Status Transition

For Transition of Payment statuses, please refer to the Credit Card Payment Service Overview.

Offering Plans

AI fraud detection offers the following plans:

Provided functions		Free Plan	Standard Plan	Advanced Plan
Scoring		〇	〇	〇
Rule Judgment		×	〇	〇
gateway		×	〇	〇
Additional Authentication (EMV 3-D Secure)		×	×	〇^※¹
AI Fraud Detection Administrative tools	Dashboard	〇	〇	〇
	Check the results	〇	〇	〇
	Rule Settings	×	△^※²	〇^※³
	simulation	×	〇	〇
	List Registration	×	〇	〇

*1: If you wish to use additional authentication (EMV 3-D Secure), you will need to apply separately.
*2: If you use the Standard Plan, you can use the template provided by our company (called the "default rule set"). You can change the values of each rule in the default rule set.
*3: When using the Advanced Plan, merchant can create their own rules.

Usage information and fraud detection timing

Usage Information
The information used in AI fraud detection is as follows.
If you use Link Type, it will be automatically obtained on our Payment screen.
If you use API type, you need to incorporate the JavaScript provided by our company into the merchant website and send the information necessary for fraud detection with JavaScript. For more information, see the AI fraud detection IF specification.

Category	Field name	Description	Link Type	API Type
Payment information	Merchant ID	ID for identifying the merchant	〇	〇
	Service ID	ID for identifying the merchant's service	〇	〇
	Credit card number^※¹		〇^※³	〇
	Expiration date of the credit card		〇^※³	〇
	Customer ID	A customer ID managed by the merchant	〇	〇
	Purchase ID	A purchase ID managed by the merchant	〇	〇
	Token^※²	The One-time token obtained	‐	〇
	Monetary amount including tax	Amount of charge with tax included	〇	〇
	Date and time of request		〇	〇
Device information	IP Address		〇^※⁴	〇^※⁵
	UserAgent		〇^※⁴	〇^※⁵
	Browser language		〇^※⁴	〇^※⁵
	Time on page		〇^※⁴	〇^※⁵
	Transition page		〇^※⁴	〇^※⁵
Unique Information^※⁶	Product ID	A product ID managed by the merchant	〇	〇
	Product name	A product name of your choice	〇	〇
	Number of Item		×	〇
	Item Category		×	〇
	Campaign Code		×	〇
	Merchant Use Field	Free column	×	〇

*1: This is Hash converted value by our company.
*2: Can be used when One-time tokens are issued.
*3: Entered by the end user on our Payment screen.
*4: This will be automatically obtained on our Payment screen.
*5: This is obtained by incorporating the JavaScript provided by our company into merchant website.
*6: Items other than Item ID are optional.

Timing of fraud detection
If you are using Link Type, the fraud detection timing will be performed after the "purchase confirmation screen".
If you are using API type, you can arbitrarily select the timing of fraud detection in merchant, but we recommend that you do it at a time when the amount is confirmed and the end user confirms the purchase, such as after the purchase confirmation screen

Cautions

Regarding use of the service
AI fraud detection is provided for the purpose of preventing the spread of fraudulent use in non-face-to-face transactions, and we do not guarantee that the detection results are accurate or complete.
It is merchant 's responsibility to decide whether to process this transaction.
Therefore, even if the detection result indicates that Payment did not indicate any risk of fraud, if a chargeback is confirmed, the Settlement amount will not be subject to settlement and will be borne by merchant.

Restriction
The Restriction and precautions for AI fraud detection are as follows.

No	Content	Restrictions and Cautions
1	Response Time Limit	If the response time exceeds 1 second, AI fraud detection does not perform scoring or rule judgment because it prioritizes Credit Card Payment processing. If it exceeds 1 second, If you are using Link Type, get Authorization.^※1 If you are using API type, an error is returned.^※2
2	Score inquiry period	365 days after score determination.
3	Maximum number of items that can be checked when checking results (orders (list))	The maximum number of inquiries is 10,000.
4	Maximum number of customer IDs that can be checked when checking results (customers (list))	The maximum number of inquiries is 1,000.
5	Limits on number of rules and lists	400KB per service unit. A good guideline is 10 rule sets and 1,000 lists per list unit.
6	Maximum number of simulation results that can be viewed	The maximum number of inquiries is 100.
7	Simulation execution period	It's different for Link Type and API type. If you are using Link Type, you can simulate Payment that occurred after November 1, 2020.^※3 If you are using API type, you can simulate Payment that occurred after the date when the JavaScript provided by our company was incorporated into merchant website.
8	Admin portal / Payment management tool Reflection time for related items	It takes some time for items related to Admin portal / Payment management tool to be available in the AI fraud detection Administration Tool. The guideline is 3 hours after the scoring judgment date and time.
9	Automatic Block	AI fraud detection means that if the total number of transactions with the same credit card in the past 1 hour exceeds 1,000, From the 1,001st transaction onwards, the rule judgment will be returned as a "block". No scoring will be performed.
10	Score details availability period	In principle, 30 days after score determination.
11	IP Address (Country)	The IP address (country) will be displayed based on the IP address obtained. If the IP address (country) is unknown, it will be displayed as blank.
12	Default rule set application status	When using for the first time, the default ruleset is not applied (not yet applied). Merchant are responsible for the application of the ruleset (including the content of the rules). (including checking and changing values). When you apply a ruleset, the rules are applied immediately.

*1: If you use a credit card (EMV 3-D Secure), EMV 3-D Secure authentication will be performed.
*2: For error codes, refer to the AI fraud detection IF specification.
*3: If it is necessary to switch to a AI fraud detection-compatible screen, it is possible to simulate Payment that occurred after the date of completion of switching to the screen.

Change of Plan
Merchant can change the plan of AI fraud detection in the middle of the subscription agreement.
If you wish to change your plan, please contact our sales representative.
Please note that if you change your plan, the operating permissions differ depending on the plan.

When to upgrade your plan
If you upgrade your plan, check whether or not you want to use AI fraud detection after upgrading the plan in the rule settings before upgrading, and set new rules as necessary.

Downgrading your plan
If you downgrade your plan, you will need to delete any rule settings you had before the downgrade and create new rule settings.

Service Suspension Due to Maintenance, etc.
AI fraud detection performs maintenance on an irregular basis. If there is a service outage due to maintenance, we will notify you in advance and stop the service. The basic rule is to give one week's notice. However, this does not apply if it is due to emergency maintenance.
In addition, if a service outage occurs due to AI fraud detection system maintenance or malfunction outside of our system, we will promptly notify merchant as soon as we know the situation and investigate the cause. We will notify merchant as soon as we are able to determine the cause, but please note that it may take some time to determine the cause if it is not caused by us.

Supplementary Information

There is no supplementary information in this document.

Definitions of Terms

The definitions of terms used in the AI fraud detection documentation are as follows:

Term	Synonyms	Definition
transaction	Payment	Interchange of money between a Merchant and an end user through a trade transaction.
Score	–	The result of calculating the risk of fraudulent use of credit Credit card information entered by end users

Introduction

Connection method

Billing method

Payment Services

Payment Services | Options

Payment management tool

Test

Notices

Service Overview