Analytical company Elliptic in cooperation with Massachusetts Institute of Technology (MIT) researched more than 200 000 transactions in Bitcoin Network for their connection with criminal activity – BitcoinDoubler2x learned.
To sort 203 769 transactions for total cost $6 mlrd, the research group involved an algorithm with machine learning. Results turned to be controversial enough: 77% of transactions weren`t classified at all, 21% were recognized being legal and just 2% – illegal. And still, researchers confidently claim that artificial intelligence can greatly increase efficiency of anti-money laundering procedure.
We remind you that alike research was conducted by Chainalysis company. And according to its data, part of Bitcoin transactions, connected with criminal, is valued in 1% in 2019. This means that generally data from Elliptic analysts doesn`t differ from research`s results, made by Chainalysis. Let us note, that in 2012 this indicator equalled 7%.
Algorithms Help to Reveal Criminal Activity
Law enforcement often turns to Elliptic for assistance, especially when it`s necessary to reveal cases of illegal cryptocurrency usage. Algorithms, developed in the company, help to determine, if BTC is used for legal purposes, for example, by persons, who have access to banking services, or if unknown intruders try to use cryptocurrency for criminal activity.
“Despite the high performance indicators of our algorithms, their usage at the current moment is associated with some troubles. The largest problem consists in false positives. The major goal of this research was to decrease the number of such positives. However, key result is that alike algorithms with machine learning usage are very effective for revealing of illegal transactions”, – the founder of Elliptic Tom Robinson claimed.
In addition, Tom Robinson noted that in some cases the system found out patterns, which are difficult to describe. But they corresponded confirmed cases of illegal activity, connected with markets on the Darknet, ransomware attacks and other criminal actions.