KEY POINTS
- Elliptic trained an AI model to identify transaction chains that represent Bitcoin laundering
- The model detected some of the known crypto laundering patterns in the sector such as peeling chains
- It also helped identify illicit digital wallets that were previously not known to the blockchain research circle
Blockchain analytics firm Elliptic has shared new details about its research on the use of artificial intelligence to detect money laundering activities in the cryptocurrency central to Bitcoin, the world's largest crypto by market cap.
The research, which was first published by Elliptic in joint studies with the MIT-IBM Watson AI Lab in 2019, has now made more significant progress, the prominent crypto forensics firm said in a blog post Wednesday.
"A machine learning model was trained to identify Bitcoin transactions made by illicit actors, such as ransomware groups or darknet marketplaces," the company said. In the updated study, Elliptic revealed how it applied new techniques to a larger data set, "containing nearly 200 million transactions."
Previously an AI model was trained to identify transactions made by illicit actors based on 200,000 transactions. This time around, a machine learning model was trained on 200 million transactions to identify "subgraphs," which Elliptic describes as "chains of transactions that represent Bitcoin being laundered." The said new AI model training approach allows for focusing on "the 'multi-hop' laundering process more generally rather than the on-chain behavior of specific illicit actors."
Elliptic's trained AI model detected some of the known crypto laundering patterns, such as peeling chains, wherein a small amount of digital assets are sent to one address while the larger chunk is sent to another address that the sender controls.
The model can also be used to help identify "previously-unknown illicit wallets," which is a significant breakthrough in the industry, considering the multiple exploits within the digital assets sector so far in the year.
"This novel work demonstrates that AI methods can be applied to blockchain data to identify illicit wallets and money laundering patterns, which were previously hidden from view," the on-chain data analytics research firm said. "Further collaboration and data-sharing will be key to advancing these techniques further and combatting financing crime in cryptoassets," it noted.
London-headquartered Elliptic's latest findings on the use of trained machine learning models to detect laundering of Bitcoins came after blockchain security firms released their data regarding the scope of exploits in the digital assets space last month.
In April alone, there was a total of $25.7 million lost to exploits, hacks and scams in the industry, with a staggering $21 million of the said figure lost to hackings. Some $4.3 million, on the other hand, were losses due to exit scams, as per data compiled by blockchain analysis and security firm CertiK.
Another blockchain security firm, PeckShield, revealed that the crypto space faced a total of 40 hacks during the month of April alone.
