Blockchain analytics firm Elliptic has announced that its researchers have made progress in detecting Bitcoin money laundering using artificial intelligence (AI).
A recent paper detailing this research was co-authored with researchers at the MIT-IBM Watson AI Lab. Elliptic reveals that it has successfully used deep learning models to identify illicit proceeds deposited on cryptocurrency exchanges, new patterns of money laundering transactions, and previously unidentified illicit wallets. did.
The company said the findings are already being used to improve its products.
Exposing hidden money laundering patterns
The underlying data is publicly available. This dataset, which boasts more than 200 million transactions, will enable the community to develop new AI methods to identify illicit cryptocurrency activity, Elliptic said in a statement this week.
Rather than identifying transactions made by criminals, the machine learning model is trained to recognize “subgraphs,” which are essentially chains of transactions that indicate Bitcoin laundering. This technique focuses on identifying these subgraphs rather than illicit wallets, allowing Elliptic to focus on the broader “multi-hop” laundering process rather than the specific on-chain behavior of individual criminals. .
Elliptic said it wiretapped a private cryptocurrency exchange to test whether the technology could detect money laundering attempts through it. Of the 52 “money laundering” subgraphs predicted to end in deposits to this exchange, the exchange confirmed that 14 were related to the reported user. On average, less than 1 in 10,000 accounts are flagged, indicating that this model performs well.
“While we have just scratched the surface of what is possible in this space, this work is already benefiting Elliptic users. Further collaboration and data sharing will continue to advance these technologies even further.” , will be the key to combating financial crime in crypto assets.”
The role of AI in the blockchain field
AI tools are increasingly showing extraordinary abilities in analyzing wide-ranging datasets to flag patterns beyond human recognition, such as identifying illicit fund movements within the Bitcoin economy .
So it's no surprise that VC investments in Web3 and AI startups exceeded $637 million in 2023.
In fact, as stated in Nansen's report, AI agents are predicted to dominate the blockchain space in 2024, establishing a more secure and efficient ecosystem.
