Skip to main content

Scorechain introduces AI for fighting cryptocurrency fraud

Publish date: 20 November 2019
Author: Scorechain

Scorechain introduces Artificial Intelligence able to detect fraud patterns in its Anti-Money Laundering cryptocurrency transaction monitoring software.

AI feature aims to provide a more accurate risk-scoring for cryptocurrencies activity with better blockchain services identification.

With Scorechain Artificial Intelligence (aka. SCai), the platform is now capable of predicting the type of service a  group of addresses stands for. The scalable model has already proved its effectiveness since SCai has given more than 92% of good predictions during the validation phase. 

SCai makes a prediction about entity types when it is related to Darkweb marketplaces, Gambling, Mixing and Exchanges platforms. The AI is based on a machine learning prediction model, the more the program is trained, the more accurate are the results it delivers. 

The SCai bot has been developed based on academic research and learns on Scorechain labeled data. The AI model has been fed with more than 90 attributes able to analyze transaction patterns and services behaviors for more than 120 million addresses so far. 

After one month of use, Scorechain Darkweb marketplace coverage has been multiplied by 3.

‘Finding new entities and improving our Blockchain naming coverage is one of our top priorities. Our clients want to have the most rigorous risk-based approach. This is a major breakthrough since the AI already identified 5 million addresses such as Darkweb marketplaces. Results have proved false positive are really low.’ said Pierre Gérard, Scorechain CEO. 

Scorechain users can decide to activate or not SCai at any time and have access to full customization. Today the AI is available for Bitcoin transaction monitoring. It follows the startup logic since every feature can be customized. They can choose when AI can be displayed depending on the percentage of certainty and which type of result it is for.  This full customization allows users to consider false positive or not and to run their own due diligence for some specific behaviors. The feature is available on the user interface and via the API.