Algorithm for automated detection
A research project has created an algorithm for the automated detection of illegal money transfers. The conditions for the use of artificial intelligence in sensitive sectors will be verifiable. Current software-based detection technologies struggle with imprecision and frequently produce inaccurate results against the laundering of illicit funds. As a consequence of this, the authorities that investigate crimes are usually overworked because they are required to examine every suspicion. Researchers are currently working on a solution that employs machine learning, a technique of artificial intelligence, to improve the search for illegal money flows and make it more precise so that fewer false alarms are generated.
Read More links for Fintech – FinTech RADAR: 105 U.S. Fintech Unicorns And Their Core Offerings
Laundering of illicit funds
The investigation and examination of monetary transactions constitute an essential part of the battle against the laundering of illicit funds. However, the existing methods of analysis uncover an excessive number of suspicious situations, each of which needs to be checked individually by analysts who have received specialized training. Only in the past two years has the competent authority, known as the Financial Intelligence Unit or FIU for short, received an average of over 300,000 reports annually, and it still has approximately 290,000 alerts that have not yet been completely handled. The use of methodologies based on artificial intelligence promises improved analysis choices, which will result in fewer false positives. In order to produce the essential technical answer, the research team is now making use of machine-learning approaches.
Read: What Is Machine Learning?
The MaLeFiz project
The new research project has the acronym MaLeFiz and it is called “Machine Learning for the Identification of Conspicuous Financial Transactions”. The MaLeFiz project will be carried out over the course of three years and will receive funding from the German federal ministry responsible for education and research. The Fraunhofer SIT is in charge of both the leadership of the project as a whole as well as the creation of the powered-by-artificial intelligence tool.
Additionally, the participants in the project are working to define minimal requirements and control mechanisms for artificial intelligence solutions that are employed in the financial industry. In addition to that, it is planned to make the outcomes of the AI traceable. Deloitte GmbH, the Fraunhofer Institute for Secure Information Technology SIT, the Martin Luther University Halle-Wittenberg, the University of Leipzig, and the Centre for Technology and Society at TU Berlin are all participating partners in this project.
Decisions that are fully transparent to interpretation by artificial intelligence
In order for results of these kind of analysis to be admissible in court, the IT solutions must fulfil a number of prerequisites. For instance, the decisions that are made by an AI need to be understandable: The artificial intelligence shouldn’t function as a “black box” that merely spits out a list of questionable cases. The criteria that it uses to determine whether or not a case is suspicious ought to be made public. In light of this, the group is currently doing study into various ethical and legal concerns. One of the objectives of the project is to compile a list of minimal requirements for artificial intelligence (AI) solutions in the financial industry.
Generate AI Images From Text Using DALL-E : A Revolutionary Tool By OpenAI
These requirements should be able to be validated, for instance, as part of an audit; they should function as a form of TÜV for apps that make use of AI. In order to find out what the AI should be like in practice and to ensure that the demands of users are taken into consideration to the greatest extent feasible, the partners in the project are conducting user interviews, user workshops, and user testing. After that, these findings will be included into a demonstrator, which will also be tested in banks in real-life settings. Lastly, the demonstrator will be evaluated. After the project wraps up in September 2025, the demonstration, the catalog of minimal needs, and other project outcomes will be made available to the general public.
[To share your insights with us, please write to sghosh@martechseries.com]
