Artificial intelligence can rapidly scan the massive Bitcoin blockchain for signs of illegal activity, a capability that could dramatically strengthen the work of law enforcement in combatting money laundering, a new report explains.
Blockchain analysis firm Elliptic on Wednesday published a study with the MIT-IBM Watson AI Lab that analyzed Bitcoin transactions using a deep learning AI model to detect money laundering patterns and identify wallets used in crimes.
Bitcoin is renowned for using a decentralized public ledger—a core facet of the technology that also made the study possible, Elliptic said. Elliptic and MIT-IBM used AI to separate legal and illicit transactions into groups and follow links among the latter to uncover potential money laundering.
“Blockchains provide fertile ground for machine learning techniques, thanks to the availability of both transaction data and information on the types of entities that are transacting, collected by us and others,” Elliptic wrote. “This is in contrast to traditional finance where transaction data is typically siloed, making it challenging to apply these techniques.”
The test was not attempted with so-called “privacy coins.”
“Counterparty information is available for all transactions [on Bitcoin],” Elliptic co-founder and Chief Scientist Tom Robinson told Decrypt, adding that the technique could be applied to other open blockchains like Solana and Ethereum. “This isn’t the case for privacy coins such as Monero.”
New Elliptic research released today explores how #AI can be leveraged to detect money laundering and other financial crime on the blockchain. The research applies new techniques to a dataset containing 200m+ transactions, which is now publicly available. https://t.co/k3GdjWJ08P
— Elliptic (@elliptic) May 1, 2024
“Rather than identifying transactions made by illicit actors, a machine learning model is trained to identify ‘subgraphs,’ chains of transactions that represent Bitcoin being laundered,” Elliptic explained. “This approach allows us to focus on the ‘multi-hop’ laundering process more generally rather than the on-chain behavior of specific illicit actors.”
According to Elliptic, the report published today continues work that the company began with MIT-IBM Watson AI Lab in 2019. Elliptic initially worked with an unnamed cryptocurrency exchange to test their theory.
“Of 52 ‘money laundering’ subgraphs predicted by the model [that] ended with deposits to this exchange, the exchange confirmed that 14 had been received by users who had already been flagged as being linked to money laundering,” the report said. “On average, less than one in 10,000 of these accounts are flagged as such, suggesting that the model performs very well.”
Government regulators, particularly in the United States, have used money laundering laws as one of many cudgels to attack the cryptocurrency industry. On Tuesday, a U.S. federal judge in Seattle sentenced Binance founder Changpeng “CZ” Zhao to four months in federal prison for money laundering violations.
Last month, the founders of the Bitcoin mixer Samourai Wallet were arrested on charges of money laundering by the U.S. Department of Justice.
In the indictment, FBI Assistant Director in Charge James Smith said “threat actors” use technology like Samourai Wallet to evade law enforcement detection and create environments conducive to criminal activity.
“While offering Samourai as a ‘privacy’ service, the defendants knew that it was a haven for criminals to engage in large-scale money laundering and sanctions evasion,” the DOJ said in the indictment. ”Indeed, as the defendants intended and well knew, a substantial portion of the funds that Samourai processed were criminal proceeds passed through Samourai for purposes of concealment.”
Edited by Ryan Ozawa.
