Elliptic's AI-driven research helps fight Bitcoin money laundering

Share this article

Blockchain analytics firm Elliptic has made significant progress in using artificial intelligence (AI) to identify potential money laundering patterns on the Bitcoin blockchain.

In a research paper co-authored with the MIT-IBM Watson AI Lab, Elliptic described its use of a deep learning model trained on a dataset of nearly 200 million transactions to detect illicit activity involving groups of bitcoin nodes and transaction chains.

The research builds upon a previous study conducted in 2019, which utilized a much smaller dataset of 200,000 transactions. The new “Elliptic2” dataset contains 122,000 labeled “subgraphs,” representing groups of connected nodes and transaction chains known to have links to illicit activity. By training the AI model on this extensive dataset, Elliptic aims to improve the accuracy and efficiency of detecting money laundering and other financial crimes on the blockchain.

The inherent transparency of blockchain data makes it well-suited for machine learning techniques, as transaction information and entity types can be readily analyzed. This stands in contrast to traditional finance, where transaction data is often siloed, making the application of AI more challenging.

The trained model successfully identified proceeds of crime deposited at a crypto exchange, as well as novel money laundering transaction patterns and previously-unknown illicit wallets. These findings are already being incorporated into Elliptic’s products to enhance their capabilities.

“The money laundering techniques identified by the model have been identified because they are prevalent in bitcoin,” Elliptic co-founder Tom Robinson explained in an email. “Crypto laundering practices will evolve over time as they cease being effective, but an advantage of an AI/deep learning approach is that new money laundering patterns are identified automatically as they emerge.”

The research revealed common money laundering techniques, such as “peeling chains,” where a user sends cryptocurrency to a destination address while sending the remainder to another address under their control, forming a chain of transactions. Another prevalent method involved the use of “nested services,” businesses that move funds through accounts at larger crypto exchanges, sometimes even without the exchange’s knowledge or approval.

To encourage further collaboration and advancement in this field, Elliptic has made the “Elliptic2” dataset publicly available. As the largest public dataset of its kind, it will enable the wider community to develop new AI techniques for detecting illicit cryptocurrency transactions and contribute to the ongoing fight against financial crime in the crypto space.

Share this article