MAS FEAT Compliance

Singapore PII Detection - Now in Community

NRIC, FIN, UEN, Phone, and Postal Code detection is now available in AxonFlow Community Edition. No license required for core Singapore PII protection.

The MAS FEAT Principles (Fairness, Ethics, Accountability, Transparency) and AI Risk Management Guidelines establish requirements for AI systems in Singapore financial services. AxonFlow provides compliance infrastructure for organizations subject to MAS regulations.

What is MAS FEAT?

MAS FEAT is a governance framework for responsible AI use in financial services, covering:

Pillar	Requirements
Fairness	Avoid unfair discrimination, test for bias, monitor demographic parity
Ethics	Human oversight, appeal processes, responsible data use
Accountability	Clear ownership, documented decisions, audit trails
Transparency	Explainable decisions, AI disclosure, customer communication

MAS AI Risk Management Guidelines (2025)

The guidelines mandate:

1. AI System Registry - Inventory of all AI systems with risk classification
2. 3-Dimensional Risk Rating - Customer Impact × Model Complexity × Human Reliance
3. Materiality Classification - High/Medium/Low based on combined risk score
4. FEAT Assessments - Periodic evaluation against all four pillars
5. Kill Switch - Emergency disable capability for AI systems
6. 7-Year Audit Retention - Complete decision logs and assessment history

Industry Applicability

MAS FEAT applies to AI systems in Singapore financial services:

Industry	Risk Level	Common Use Cases
Banking	High	Credit scoring, fraud detection, AML
Insurance	High	Underwriting, claims processing
Capital Markets	High	Trading algorithms, robo-advisory
Payments	Medium-High	Fraud detection, transaction monitoring

Community vs Enterprise Features

Feature	Community	Enterprise
Audit logging	✅	✅
Policy enforcement	✅	✅
Standard PII detection (SSN, credit cards)	✅	✅
Singapore NRIC detection	✅	✅
Singapore FIN detection	✅	✅
Singapore UEN detection	✅	✅
Singapore phone detection	✅	✅
Singapore postal code detection	✅	✅
NRIC/FIN checksum validation		✅
AI System Registry		✅
FEAT Assessments		✅
Kill Switch		✅
7-Year Retention		✅
MAS Export		✅
Compliance Dashboard		✅

Singapore PII Detection (Community)

Community edition includes detection for all Singapore PII types:

NRIC Detection

National Registration Identity Card - 9-character alphanumeric identifier:

Format: S1234567D
        ^       ^
        |       +-- Check letter (A-Z)
        +---------- Prefix letter
                    S = Citizen born before 2000
                    T = Citizen born 2000+
                    M = Foreigner from 2022+

Detected patterns:

• S1234567D - Citizen (pre-2000)
• T9876543J - Citizen (2000+)
• M1234567K - Foreigner (2022+)

FIN Detection

Foreign Identification Number - same format as NRIC but for foreigners:

Format: F1234567N
        ^       ^
        |       +-- Check letter (A-Z)
        +---------- Prefix letter
                    F = Foreigner pre-2000
                    G = Foreigner 2000-2021

UEN Detection

Unique Entity Number - business registration identifier:

Formats:
- 53276128A     (8-digit: old business registration)
- 200312345A    (9-digit: company registration YYYYNNNNNC)

Phone Detection

Singapore phone numbers with +65 country code:

Formats:
+65 9123 4567   (Mobile - starts with 8 or 9)
+65 6234 5678   (Landline - starts with 6)

Postal Code Detection

6-digit Singapore postal codes (logged but not blocked):

Format: Singapore 238877

Detection Behavior

Pattern	Request	Response	Severity
NRIC	Warn	Redact	Critical
FIN	Warn	Redact	Critical
UEN	Warn	Redact	High
Phone	Warn	Redact	Medium
Postal	Warn	Log only	Low

SDK Examples

Python

from axonflow import AxonFlow

async with AxonFlow() as client:
    result = await client.get_policy_approved_context(
        user_token="sg-user",
        query="Customer NRIC is S1234567D"
    )
    print(f"Policies: {result.policies}")  # ['sys_pii_singapore_nric']

TypeScript

import { AxonFlow } from '@axonflow/sdk';

const client = new AxonFlow();
const result = await client.getPolicyApprovedContext({
  userToken: "sg-user",
  query: "Customer NRIC is S1234567D"
});
console.log(result.policies);  // ['sys_pii_singapore_nric']

Go

client := axonflow.New()
result, _ := client.GetPolicyApprovedContext(ctx, &axonflow.PolicyApprovalRequest{
    UserToken: "sg-user",
    Query:     "Customer NRIC is S1234567D",
})
fmt.Println(result.Policies) // [sys_pii_singapore_nric]

Java

AxonFlow client = AxonFlow.create();
PolicyApprovalResult result = client.getPolicyApprovedContext(
    PolicyApprovalRequest.builder()
        .userToken("sg-user")
        .query("Customer NRIC is S1234567D")
        .build()
);
System.out.println(result.getPolicies()); // [sys_pii_singapore_nric]

Enterprise Features

NRIC/FIN Checksum Validation

Enterprise includes mathematical checksum validation:

S1234567D  →  Valid (checksum D verified)
S1234567A  →  Invalid (checksum should be D)

Reduces false positives by rejecting invalid IDs.

AI System Registry (Enterprise)

Enterprise provides MAS-compliant AI Model Registry:

• Model versioning and approval tracking
• Risk classification (Tier 1-3)
• Change management logging
• Regulator-accessible audit trails

3-Dimensional Risk Rating (Enterprise)

Enterprise automatically calculates materiality based on:

Dimension	Scale	Description
Customer Impact	1-5	Potential harm to customers
Model Complexity	1-5	Technical sophistication
Human Reliance	1-5	Degree of automation vs oversight

Materiality Classification:

Sum of Ratings	Classification	Assessment Frequency
>= 12	High	Quarterly
>= 8	Medium	Semi-annually
< 8	Low	Annually

Quick Start

Test Singapore PII Detection (Community)

Singapore PII detection is available now in Community Edition:

# Start AxonFlow
git clone https://github.com/getaxonflow/axonflow.git
cd axonflow
docker compose up -d

# Test NRIC detection
curl -X POST http://localhost:8080/api/policy/pre-check \
  -H "Content-Type: application/json" \
  -d '{"query": "Customer NRIC is S1234567D", "client_id": "test"}'

Expected response:

{
  "approved": true,
  "policies": ["sys_pii_singapore_nric"],
  "requires_redaction": true
}

Run Example Tests

cd examples/singapore-pii/python
pip install axonflow
python demo.py  # 12 tests covering all patterns

Implementation Checklist

Community (Available Now)

• Deploy AxonFlow (Docker Compose or Kubernetes)
• Singapore PII detection enabled by default
• Configure SDK in your application
• Review audit logs for PII detections
• Test with Singapore PII patterns

Enterprise (Contact Sales)

• Enable NRIC/FIN checksum validation
• Configure AI System Registry
• Set up FEAT assessments
• Configure 7-year retention
• Enable compliance dashboard
• Set up HITL for high-risk decisions

Related Regulations

MAS FEAT compliance features also help with:

• PDPA Singapore - Personal data protection
• Technology Risk Management Guidelines - MAS technology governance
• Business Continuity Management - Kill switch aligns with BCM requirements
• SEBI AI/ML Guidelines - Similar requirements for India (View)
• EU AI Act - European AI regulations (View)

Resources

• MAS FEAT Principles (PDF)
• MAS AI Risk Management Guidelines
• PDPA Singapore
• Singapore PII Examples

Get Enterprise Features

For full MAS FEAT compliance capabilities:

• Request a Demo - See AI System Registry and FEAT assessments
• Contact Sales - Discuss your compliance requirements

Enterprise Documentation

Enterprise customers: See the MAS FEAT Compliance Guide for detailed API documentation, SDK examples, and configuration options.