- Kyung Hee University professor Kyoung Jun LEE emphasized that "Korea should strive to become a leading country in the Asian financial crime prevention network."
Advancing Anti-Money Laundering with AI: Experts Highlight the Role of Federated Learning
The "User Centric Artificial Intelligence (UCAI) Forum Seminar" concluded successfully on June 19 at the National Assembly Hall. The seminar, centered around the theme of "Korea-U.S. Artificial Intelligence(AI) Cooperation to Prevent Financial Crimes and Strengthen Competitiveness," was co-hosted by South Korean lawmakers Sungwon Kim, the Korea Management Information Society, and the US's McLarty Associates. This event was held as part of the 3rd UCAI Forum and featured congratulatory speeches from Vice Chairman of the National Assembly Woo-taik Chung, former U.S. Ambassador to South Korea Thomas Hubbard, and lawmaker Seok-joon Song
In his opening remarks, Kyung Yang PARK, Co-Chair of the UCAI Forum, noted the inefficiency of the existing system, which spends a staggering $270 billion every year to detect financial crimes, but has a detection rate of only 1%, and said that through this seminar, he hopes that Korea and the United States will create a global best practice by utilizing the most advanced UCAI shared method of federated learning AI technology to prevent financial crimes and strengthen financial competitiveness.
Lawmaker Sung-won Kim said in a welcoming speech that South Korea's Yoon Suk Yeol government is taking the lead in accelerating AI-based innovation and preparing measures to revitalize AI that suit the characteristics of the financial sector. He also thanked everyone who attended the discussion on the various uses of AI in the field of financial crime prevention.
Woo-taek Chung, vice chairman of the National Assembly, said in a congratulatory speech, "AI technology is steadily developing, but security systems have a detection rate of less than 1%, and technology sharing should also be carried out in the financial security technology field. "He also said he expects the seminar to actively discuss how the Korea-U.S. AI technology alliance will develop the future of the financial security technology sector.
Thomas Hubbard, former U.S. ambassador to South Korea, said in a congratulatory speech that he is happy to be here to prevent financial crimes and strengthen competitiveness through the U.S.-Korea alliance. He said that the U.S.-Korea AI technology alliance between Korea's Harex Infotech and the U.S. Consilient, which was signed during President Yoon Suk Yeol's visit to the U.S. in April 2023, is very important in terms of promoting cooperation to solve important problems through technology development in all fields of the private and public sectors. He also said that he expects AI to play a big role in practical issues such as anti-money laundering(AML) and terrorist financing, and said that UCAI will play an important role in humanity.
Seok-Joon Song, a member of the National Assembly, said In his congratulatory remarks, "Congratulations on Korea and the United States having AI cooperation seminars in a meaningful year marking the 70th anniversary of the Korea-U.S. alliance." It is also very meaningful for top experts from South Korea and the United States to gather to discuss AI and financial crime prevention, and the National Assembly also said it will participate in efforts such as systems and legislation to prevent financial crimes through AI.
Jong-won Kim, chairman of the Korea Management Information Science Association, said in a congratulatory speech, "Currently, financial crimes are emerging as a serious problem around the world." Financial crimes have a negative impact on society as a whole, such as threatening the stability of the financial system, and artificial intelligence will play an important role in preventing financial crimes, strengthening financial system stability, and enhancing the competitiveness of financial services. The seminar said today that it is hoped that the artificial intelligence technology alliance between South Korea and the U.S. will serve as an opportunity for cooperation to enhance the competitiveness of the nation's financial industry by increasing the detection rate of financial crimes based on federated learning.
In the following main session, Professor Sang-mi Chae at Ewha Womans University, and a research vice chairman of the Korea Management Information Society, presented on the topic of "Machine Learning and AI for Detecting Digital Asset Abnormal Transactions and Autonomous Response." Professor Chae highlighted the severity of financial crimes in the domestic market by introducing abnormal financial transactions in the digital asset market and cases of organized market price manipulation in Korea. He announced the successful detection performance of machine learning and AI technologies in laying the foundation for measures to prevent fraudulent activities known as "pump and dump" in the crypto asset market.
Afterward, Min-Wook Kang, Team Leader of SC First Bank, presented "The Movement of Illegal Funds in Financial Institutions and the Application of Machine Learning Technology." Kang provided an overview of financial crime in the domestic financial industry, including money laundering, the movement of illegal funds in financial institutions, and measures to prevent such activities. Kang emphasized that while the adoption of machine learning and AI may initially incur costs, it offers significant advantages in terms of operational efficiency and cost-effectiveness in the long run. He also highlighted the difficulty of timely dealing with financial crimes due to insufficient manpower and stressed the need to consider technology-based approaches to significantly improve the response level to financial crimes.
Gary Shiffman, CEO of Consilient delivers a presentation on "UCAI, How to Achieve the Ultimate Goal of Federated Learning"/ Courtesy of UCAI Forum
Next, Gary Shiffman, CEO of Consilient, delivered a presentation on "UCAI, How to Achieve the Ultimate Goal of Federated Learning," focusing on the Korea-U.S. AI Technology Alliance, a key theme of the seminar. Schiffman proposed At the "Global SMEs Night" event held in Washington, DC, on April 25 during South Korean President Yoon Suk Yeol's state visit to the US, Minister of SMEs and Startups Young Lee had an in-depth conversation with Kyung Yang PARK, President of Harex Infotech, a Korean company leading a User Centric AI Sharing Platform, and Juan Zarate and Gary Shiffman, experts of financial crime from Consultant of Consilient. The Consilient was co-founded in 2020 to lead the way in the deployment of AML and countering the financing of terrorism (CFT) systems by leveraging AI and Federated learning to prevent financial crime. Juan Zarate served as U.S. Deputy National Security Advisor and Assistant Secretary of State for Terrorism and Finance, and Gary Schiffman was a national security expert in the U.S. Senate and Homeland Security. Zarate led several research programs funded by the Defense Advanced Research Projects Agency (DARPA) and is the founder of Giant Oak and the inventor of GOST, a machine learning-based screening platform.
During the forum, Schiffman presented the world's first financial crime detection results through federated learning, demonstrating that the number of financial crime risk notifications required per month could be reduced from 800 to 170 by one employee in the AML department. He stated, "We observed six times faster identification of crime-related financial cases."
In a panel discussion moderated by Sang-man Han, President of Sungkyunkwan University, Professor Kyoung Jun LEE from Kyung Hee University emphasized the importance of introducing AI into Korea's Financial Intelligence Agency system under the Financial Services Commission to effectively prevent financial crimes. He explained that the existing system and operation of the Financial Services Commission do not satisfy the requirements for combating financial crimes. While these systems may inconvenience criminal groups seeking short-term illegal gains and some financial institution employees overly focused on short-term performance, they contribute to strengthening the country, and the market economy, and driving national development in the medium to long term.
Professor Lee proposed measures to enhance cooperation and competition among banks, securities, insurance, and cryptocurrency exchanges, along with the advancement of financial crime prevention technology. He stated, "To advance financial crime prevention technology, we must first encourage the adoption of machine learning and deep learning-based systems for financial crime detection to simultaneously improve performance and efficiency."
Addressing concerns about data sharing, Professor Lee explained that joint learning is a proven technology used in various sectors, such as medicine, pharmaceuticals, manufacturing, and finance. It enables economic players to build robust AI systems without sharing sensitive data. He also emphasized that Korea should join the financial crime prevention network and become a leading country in Asia. He stated, "We need to establish an AI sharing platform that connects not only financial institutions but also customs offices, tax offices, and immigration offices."
Professor Seokjin Hwang of Dongguk University then emphasized the importance of AI technology based on federated learning in the field of AML. He explained that methods and tools for money laundering are gradually evolving, and the existing AML system relies heavily on monitoring personnel. To solve this problem, AI will be able to solve various problems by analyzing usage patterns and alleviating reliance on monitoring personnel, and it is necessary to upgrade the AML system using AI.
Jae-Wook Jung, a partner at Joowon Law Firm, cited the possibility of "endogenous errors" as a factor that needs to be considered when utilizing AI. 'Endogenous error' refers to problems caused by the accuracy and reliability of the data on which the AI algorithm is based. He also mentioned the problem of liability and the possibility of discrimination by AI and said that it is necessary to think about a methodology that can solve these problems.
Ki-soo Shin, head of the Financial Data Policy Division of the Financial Services Commission, explained the AI activation policy being carried out by the financial authorities. The Financial Services Commission announced that it produced AI guidelines that were used in five areas: transaction detection, chatbots, recommendation services, and credit ratings in 2021. In addition, in August 2022, it said that it prepared a plan to activate AI and secure reliability, including securing high-quality data, establishing a data neutrality environment, resolving regulatory and institutional uncertainties, and establishing a trusted environment. In the case of abnormal transaction detection, Shin said, "The amount of data is insufficient compared to normal transactions, and it is difficult to share and open data," adding, "The federated learning method was very impressive as a way to solve the problem of personal information protection while efficiently securing insufficient data." expressed one's feelings.
Korea-U.S. AI Technology Alliance: Collaborative Efforts to Prevent Financial Crimes through Innovative AI Solutions
At the "Global SMEs Night" event held in Washington, DC, on April 25 during South Korean President Yoon Seok-yeol's state visit to the US, Minister of SMEs and Startups Young Lee had an in-depth conversation with Kyung Yang Park, President of Harex Infotech, a Korean company leading a User Centric AI Sharing Platform, and Juan Zarate and Gary Shiffman, experts of financial crime from Consultant of Consilient. The Consilient was co-founded in 2020 to lead the way in the deployment of AML and countering the financing of terrorism (CFT) systems by leveraging AI and federated learning to prevent financial crime. Juan Zarate served as U.S. Deputy National Security Advisor and Assistant Secretary of State for Terrorism and Finance, and Gary Schiffman was a national security expert in the U.S. Senate and Homeland Security. Zarate led several research programs funded by the Defense Advanced Research Projects Agency (DARPA) and is the founder of Giant Oak and the inventor of GOST, a machine learning-based screening platform.
Consilient's solution optimizes the next-generation machine learning model through federated learning, increasing the risk detection rate of financial crime and protecting personal information while protecting the financial system. This innovative network approach allows organizations to share optimized models without moving or extracting data, facilitating continuous learning, discovery of unseen risks, and evolution of AML strategies.
As part of the detailed implementation plan for the Korea-U.S. AI Technology Alliance, Consilient will collaborate with Harex Infotech to build a global model for financial crime prevention by applying Consultant solutions to Korean financial institutions. This collaboration is expected to strengthen the financial competitiveness of both companies and become a new growth engine globally, including South Korea's next-generation advancements.
