Skip to main content
Back to challenges
Challenge

Automate Synthetic Identity Fraud Triage with AutoGen Multi-Agent Systems

Synthetic identity fraud combines real and fabricated PII to bypass traditional credit checks. Construct an AutoGen multi-agent system that analyzes social graph linkages and bureau data to isolate synthetic identity networks with 95% detection accuracy.

CybersecurityHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build an AutoGen multi-agent team to cross-examine identity signals and social graphs for synthetic identity fraud detection.

Delivery guide

How work is evaluated

Evaluation

Evaluates synthetic identity detection accuracy across clusters of fabricated credit profiles.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

Sleeper Identity Detection Test

Fabricated identities with high shared address usage must flag is_synthetic == True.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

Synthetic Fraud Recall

Ratio of correctly isolated synthetic identity profiles. • target: 0.95 • range: 0.9-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Configure AutoGen agents representing PII Verifier, Graph Link Analyst, and Fraud Supervisor.

  • Perform cross-entity resolution over shared SSNs, phone numbers, and delivery addresses.

  • Detect identity fabrication patterns such as sleeper profiles and credit-padders.

  • Generate prioritized investigator queues with graphical evidence trees.

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic credit applicant dataset containing embedded Frankenstein identity networks.

Challenge setup

Evaluates synthetic identity detection accuracy across clusters of fabricated credit profiles.

Challenge input

JSON containing ssn, name, address, shared_phone_count, account_age_days

AutoGen

Orchestrates multi-perspective agent conversations for identity verification.

Evaluated output

JSON with is_synthetic (boolean), synthetic_score (0-100), evidence_summary

Evaluation checks
  • Fabricated identities with high shared address usage must flag is_synthetic == True.
  • Ratio of correctly isolated synthetic identity profiles. • target: 0.95 • range: 0.9-1
Configured evaluation evidence
  • Benchmark: SyntheticIdentityBench
  • Synthetic Fraud Recall target: 0.95
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • AutoGen · Required
  • Microsoft AutoGen · Optional

Evaluation contract

  • Sleeper Identity Detection Test · Weight 1
  • Synthetic Fraud Recall · Weight 1

Metrics

  • Synthetic Fraud Recall · Target 0.95

Benchmarks

  • SyntheticIdentityBench
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing ssn, name, address, shared_phone_count, account_age_days

    Output: JSON with is_synthetic (boolean), synthetic_score (0-100), evidence_summary

    Execution harness: Python

Evaluation tools

  • AutoGen · Agent Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Automate Synthetic Identity Fraud Triage with AutoGen Multi-Agent Systems