In Short:
A new study from Elliptic, MIT, and IBM has developed an AI tool to detect money laundering on Bitcoin’s blockchain. This tool uses patterns of transactions leading from criminal entities to exchanges where dirty crypto is cashed out. The researchers have released a vast data set of 200 million transactions to train the AI model, marking a significant advancement in automated blockchain analytics.
New Study Enhances Blockchain Analytics through AI
Artificial intelligence (AI) tools have been exceptionally effective in analyzing large amounts of data to uncover patterns that may not be visible to humans. This has made Bitcoin’s blockchain, which records nearly a billion transactions between pseudonymous addresses, an ideal challenge for AI to tackle. A recent study, coupled with a massive release of cryptocurrency crime training data, is poised to advance automated tools in detecting illicit money flows within the Bitcoin ecosystem.
Research Findings
Researchers from cryptocurrency tracing firm Elliptic, MIT, and IBM recently published a paper outlining a new method for identifying money laundering activities on Bitcoin’s blockchain. Instead of focusing on pinpointing specific cryptocurrency wallets linked to criminal entities such as dark web markets or scammers, the researchers analyzed patterns of bitcoin transactions leading from known bad actors to cryptocurrency exchanges where illicit funds could be cashed out. They used these patterns to train an AI model capable of recognizing similar money movement behaviors.
The team not only introduced an experimental version of the AI model for detecting bitcoin money laundering but also shared the training dataset behind it—a 200-million transaction dataset from Elliptic. This dataset is described as the largest of its kind ever made public, offering a significant enhancement in blockchain analytics.
Significance of the Research
Elliptic, in collaboration with MIT and IBM, has been utilizing machine learning tools for years to automate the process of tracing crypto funds and identifying criminal actors. Their latest research marks a shift in the approach, focusing on patterns of money laundering behavior across the blockchain rather than individual transactions.
By analyzing collections of transactions between illicit actors and exchanges, Elliptic created a dataset of 122,000 subgraphs as examples of money laundering behavior within a total of 200 million transactions. This dataset was used to develop an AI model capable of identifying money laundering patterns throughout Bitcoin’s blockchain.
