An AI model inspired by ants, bees, and birds could be the key to cracking down on blockchain-based money laundering, new research led by Charles Darwin University (CDU) has found.
The anonymous, fast-moving, and cross-border nature of blockchain transactions has led to an increase in money laundering and terrorist financing risks globally, but the static, rule-based nature of traditional anti-money laundering (AML) detection systems are unable to keep up.
In the study, CDU Lecturer in Engineering and IT Dr Reem Sherif proposed a swarm-based agentic AI architecture, making the identification of illicit cash flows more scalable, auditable, and resilient.
Inspired by social species such as ants, the model leans on concepts such as decentralised coordination and simple rules to produce complex behaviour.
For example, ants' social hierarchy is broken into a queen, workers, soldiers, and males, each of which has a simple, singular role to further the complex goals of the colony. Similarly, one AI model might complete the role of the 'worker', while another would be the 'soldier' - each model completes a specific task to create the complex system needed to track and expose money laundering.
Dr Sherif said financial criminals were continually adapting their methods, but the proposed framework explored how a team of five specialised AI agents work together to detect suspicious transaction patterns more effectively than traditional approaches.
"Rather than relying on a single AI model, our framework assigns different AI agents to specific tasks, such as analysing transaction flows, modelling relationships between accounts, and tracking suspicious activity," she said.
"By combining multi-agent artificial intelligence with blockchain technology, we aim to improve both the accuracy and explainability of automated money laundering detection, helping financial institutions identify illicit activities while maintaining a clear audit trail.
"Together, they provide a more adaptive and transparent approach to anti-money laundering."
Moving forward, Dr Sherif said it was crucial to strengthen the swarm-based model's integration with compliance tooling.
She also said refining inter-agent protocols would better support broader jurisdictional interoperability.
"As digital financial systems continue to evolve, intelligent collaborative AI systems have the potential to strengthen regulatory compliance and support faster, more reliable detection of financial crime," she said.
Towards Agentic AI Swarm Modeling for Blockchain-Based Money Laundering Detection was published in Communications in Computer and Information Science, with contributions from CDU's Dr Mukhtar Hussain and Professor Sami Azam, as well as Professor Ernest Foo of Griffith University and Dr Zahra Jadidi of Queensland University of Technology.
