A collaborative effort between researchers from Elliptic, IBM Watson, and MIT has leveraged artificial intelligence (AI) to detect instances of money laundering on the Bitcoin blockchain.
Unveiling the Findings
In 2019, Elliptic, in conjunction with the MIT-IBM Watson AI Lab, unveiled a study showcasing how machine learning models could be trained to identify Bitcoin transactions associated with illicit actors, such as ransomware groups or dark web marketplaces.
Expanding the Scope
Building upon their prior work, the research partners have now unveiled groundbreaking research applying advanced techniques to a significantly larger dataset, comprising nearly 200 million transactions. Rather than solely focusing on transactions executed by illicit actors, the researchers trained a machine learning model to pinpoint “subgraphs” – chains of transactions indicative of Bitcoin laundering.
Shift in Focus
By targeting these subgraphs instead of specific illicit wallets, the researchers could delve into the broader “multi-hop” money laundering process. This shift enabled a more comprehensive analysis, transcending the on-chain behavior of individual actors.
Validation through Testing
In collaboration with a cryptocurrency exchange, the researchers validated their methodology. Out of 52 identified money laundering subgraphs culminating in deposits to the exchange, 14 were traced back to users previously flagged for involvement in money laundering activities.
Demonstrated Efficacy
The findings indicate a high level of efficacy, with less than one in 10,000 accounts being flagged using this approach. This statistical significance underscores the robustness of the model, implying its effectiveness in identifying potential instances of financial misconduct.
Transparency and Accessibility
Emphasizing the significance of their work, Elliptic asserts that this research underscores the potential of AI methods in uncovering illicit wallets and money laundering patterns previously concealed within blockchain data. The transparency inherent in blockchains facilitates such analyses, positioning cryptoassets as more amenable to AI-based financial crime detection compared to traditional financial instruments.
Conclusion
The collaborative efforts of Elliptic, IBM Watson, and MIT have unveiled a groundbreaking application of AI in blockchain analytics, shedding light on previously obscured money laundering patterns. Through innovative methodologies and leveraging the transparency of blockchains, this research underscores the potential of AI in combatting financial crime within the cryptoasset ecosystem.