Minority Mindset

Smart investments

Crypto News

Blockchain Sleuth Elliptic Explores AI and Anti-Money Laundering Using 200M Bitcoin Transactions

Apr 30, 2024
  • The Elliptic2 dataset is orders of magnitude bigger than the one used when the team began using machine learning to detect money laundering with bitcoin back in 2019.
  • The research made use of 122,000 groups of connected nodes and chains of transactions called “subgraphs” with known links to illicit activity.
As a researcher with experience in blockchain analytics and money laundering detection using machine learning, I find the recent work by Elliptic on detecting money laundering patterns on the Bitcoin blockchain using their new “Elliptic2” dataset of 200 million transactions to be quite intriguing. This is a significant expansion from the dataset used in their earlier program in 2019, which only included 200,000 transactions.Elliptic, a leading blockchain analysis firm, revealed that their AI system identified suspected money laundering activities on the Bitcoin network based on its learning from over 200 million past transactions.

In 2019, we conducted a study utilizing a transaction dataset comprising 200,000 records. Our latest project builds upon that research, employing the considerably larger “Elliptic2” dataset. This new dataset consists of over 122,000 identified “subgraphs,” which are clusters of interconnected nodes and chains of transactions previously flagged for their involvement in illicit activities.

As the amount of data grows larger for training machine-learning algorithms in AI, its ability to gain deeper insights becomes more profound. Transparent transaction data from cryptocurrencies such as bitcoin offer a wealth of material for this purpose. In a collaborative study with researchers from MIT-IBM Watson AI Lab, Elliptic utilized these transactions to identify the distinctive patterns associated with money laundering in cryptocurrency and effectively categorize new illicit activities.
Tom Robinson, the co-founder of Elliptic, stated via email that the money laundering methods detected by their model have been recognized due to their prevalence in cryptocurrencies such as bitcoin. He added that crypto money laundering tactics will adapt as they become less effective, but a benefit of utilizing AI/deep learning is the identification of emerging money laundering trends autonomously.

As an analyst, I’ve discovered that several questionable subgraphs consist of what are infamously called “peeling chains.” In these scenarios, a user transfers cryptocurrency to a specific destination address, but keeps the change or remainder for themselves. This process is repeated multiple times to create an intricate chain.

“According to Ellptic’s research paper, in conventional finance, large sums of money are broken down into numerous smaller transactions to bypass regulatory thresholds and evade scrutiny. This practice is referred to as ‘smurfing.'”

As a crypto investor, I’ve come across the practice of using “intermediary services” or “nested platforms” in my transactions. These are businesses that facilitate fund transfers between accounts on larger cryptocurrency exchanges, often without the explicit consent of the exchange itself.

As a financial analyst, I’ve discovered that nested services, which are integrated into larger platforms, often have less rigorous customer due diligence procedures than the cryptocurrency exchanges they rely on. In some cases, these nested services don’t even conduct any anti-money laundering checks at all. This lack of scrutiny makes them attractive targets for criminals looking to launder cryptocurrencies, which may lead to their inclusion in suspicious subgraphs according to the model I’ve been analyzing.

Read More

2024-05-01 16:20