Building a Graph-Based AML Anti-Collusion System with Didit and Neo4j

Graph Databases for AMLGraph databases like Neo4j are powerful tools for uncovering complex, non-obvious relationships between entities, crucial for detecting sophisticated AML fraud and collusion networks.

The Challenge of CollusionTraditional AML systems often struggle to identify collusion and synthetic identity fraud because they analyze transactions and identities in isolation, missing the interconnected web of malicious activity.

Leveraging Enriched Identity DataIntegrating high-quality, verified identity data from platforms like Didit is fundamental for populating a robust graph database, providing the foundational nodes for network analysis.

Didit's Role in Anti-CollusionDidit's modular identity verification solutions, including ID Verification, AML Screening, and Phone & Email Verification, provide the rich, structured data necessary to build and power effective graph-based anti-collusion systems.

The Rising Threat of Collusion and Synthetic Identities in AML

Financial crime is not static; it constantly evolves. One of the most insidious forms of fraud is collusion, where multiple individuals or entities work together to bypass anti-money laundering (AML) controls. This often involves the use of synthetic identities—fabricated personas created by combining real and fake information to open accounts, secure loans, and launder money. Traditional, rule-based AML systems, which typically analyze individual transactions or customer profiles in isolation, are often ill-equipped to detect these complex, interconnected fraud schemes. They miss the subtle patterns and shared attributes that link seemingly disparate accounts back to a single fraud ring.

The challenge lies in moving beyond simple point-in-time checks to understanding the relationships and behaviors across a network of entities. This is where graph databases, coupled with robust identity verification data, become indispensable. By mapping out connections between customers, accounts, devices, and transaction patterns, organizations can reveal hidden collusion networks that would otherwise go unnoticed.

Why Graph Databases are Essential for Anti-Collusion

Graph databases, such as Neo4j, are purpose-built to store and traverse relationships between data points efficiently. Unlike relational databases that require complex joins to infer connections, graph databases represent data as nodes (entities) and edges (relationships), making it incredibly intuitive and performant to query and visualize networks. This native ability to handle relationships is precisely what's needed for an effective anti-collusion system.

Consider a scenario where multiple customers open accounts using different names but share the same address, phone number, or even the same device during onboarding. A relational database might flag these as individual anomalies, but a graph database can immediately show them as connected nodes, potentially indicating a single fraudster or a colluding group. By linking data points like addresses, phone numbers, IP addresses, email accounts, and even biometric identifiers (derived from a 1:1 Face Match or Face Search), a graph database can expose the intricate web of relationships that define a collusion ring or a synthetic identity.

Building Your Graph-Based AML System with Didit Data

The foundation of any powerful graph-based AML system is accurate and comprehensive identity data. This is where Didit, an AI-native identity platform, plays a crucial role. Didit's modular architecture allows businesses to collect and verify a wide array of identity attributes, which then become the nodes and properties in your Neo4j graph.

Here's how Didit's products feed into a graph-based anti-collusion system:

• ID Verification (OCR, MRZ, barcodes, NFC Verification): When a user undergoes ID Verification, Didit extracts and verifies critical information from their government-issued document. This includes name, date of birth, address, document number, and issuing country. This data forms the core identity nodes in your graph. For high-security scenarios, NFC Verification of ePassports/eIDs provides cryptographic assurance of document authenticity.
• Passive & Active Liveness: Liveness detection ensures the person presenting the ID is physically present and not a deepfake or spoof. This adds a layer of trust to the identity node, reducing the risk of synthetic identities at the point of onboarding.
• 1:1 Face Match: Comparing the selfie to the ID document photo confirms the person's identity. If multiple accounts are linked to the same face but different names, this is a strong indicator of synthetic identity or collusion.
• Proof of Address: Verifying a user's address provides another crucial data point for linking identities. Multiple users at the same address, especially if combined with other shared attributes, can highlight collusion.
• Phone & Email Verification: These checks confirm contact information. Shared phone numbers or email addresses across different user accounts are classic signs of collusion.
• AML Screening & Monitoring: Didit's AML Screening screens users against 1300+ global sanctions, PEP, and watchlist databases. While providing a direct compliance outcome, the underlying match data (e.g., a partial match on a watchlist) can be ingested into the graph to identify indirect connections or associations with high-risk entities, further enriching your risk profile. The two-score system (Match Score and Risk Score) provides granular data for analysis.

Each piece of verified data from Didit becomes a node or an attribute of a node in your Neo4j graph. For example, a 'Person' node can be connected to 'Address' nodes, 'Phone Number' nodes, 'Email' nodes, and 'Device' nodes (via IP analysis or device intelligence). When a new user is onboarded, their verified data is added to the graph, and the system can immediately query for existing connections. Are there other users sharing this address? Has this phone number been used with other identities? Is this device associated with any flagged accounts?

Analyzing Graph Data for Anti-Collusion and Fraud Detection

Once your Didit-verified data is in Neo4j, you can leverage graph algorithms and queries to detect patterns indicative of collusion and synthetic identities:

• Pathfinding Algorithms: Find the shortest path between two seemingly unrelated entities. If a short path exists through shared attributes (e.g., same IP, same address, same phone), it could indicate a collusive relationship.
• Community Detection: Identify clusters of highly interconnected nodes. These communities might represent fraud rings operating together.
• Centrality Algorithms: Identify highly influential nodes (e.g., a phone number or an address that connects many different identities), which could be a central point for a fraud operation.
• Pattern Matching: Define specific fraud patterns (e.g., "multiple accounts created within a short time, sharing the same device ID but different ID documents") and query the graph to find instances.

By continuously feeding validated identity data from Didit into your Neo4j graph, you create a dynamic, self-learning system that can evolve with fraud tactics. The modularity of Didit means you can start with essential verification steps and add more sophisticated checks as your needs grow, all while ensuring your graph database is populated with the highest quality, AI-native identity data.

How Didit Helps

Didit provides the essential building blocks for a robust, graph-based anti-collusion system. Our AI-native platform offers a comprehensive suite of identity verification tools, including ID Verification, Passive & Active Liveness, 1:1 Face Match, AML Screening & Monitoring, Proof of Address, and Phone & Email Verification. These tools deliver high-quality, structured identity data crucial for populating your Neo4j graph. Didit’s modular architecture means you can select the exact verification primitives you need, ensuring you collect only relevant data while maintaining flexibility. With our Free Core KYC offering and no setup fees, you can quickly implement foundational verification steps and begin building your anti-collusion network without significant upfront investment. Our developer-first approach, with clean APIs and an instant sandbox, makes integration seamless, allowing you to focus on leveraging the power of graph analytics rather than managing complex identity infrastructure.

Ready to Get Started?

Ready to see Didit in action? Get a free demo today.

Start verifying identities for free with Didit's free tier.