SupTech Week 2024

Artificial Intelligence (AI) and Machine Learning (ML) are rapidly transforming the suptech landscape, offering powerful tools for enhanced supervision. “The real challenge is not just understanding AI but knowing how to leverage it thoughtfully and with human judgment, ensuring we are not overtaken by the technology but remain in control of its use.” as outlined by Francesca Hopwood-Road, Head of London Centre, BIS Innovation Hub when discussing future-proofing supervision. While technology plays a crucial role, the human element remains essential for effective supervision. As Minerva Tantoco, Former first CTO, City of New York and CEO, City Strategies eloquently shared ‘In our rush to embrace artificial intelligence, we often forget that at its core, AI is a tool created by humans for humans. It’s not a replacement for human intelligence but an augmentation of it. It’s humans that make technology better by making sure that we use it to serve us and not the other way around.’  In essence, while technology is a critical enabler, it is essential to recognise that the human workforce is the key to the successful adoption and integration of suptech tools. Technology alone is not enough; the skills, expertise, and engagement of staff are vital for making suptech solutions effective and efficient.

Training and skill development for staff are essential to ensure they can effectively utilise new technologies, understand the insights they provide, and make informed decisions based on them. A session on unlocking skills and personal growth discussed resources specifically available to support women in their leadership driving digital transformations across organisations.  For example, Ramūnas Baravykas, Chief Data Officer, Bank of Lithuania emphasised that supervisors need to acquire new skills and adopt new strategies to effectively navigate the data-rich environment and to use suptech tools effectively. The need for upskilling is not just about learning new tools but also about changing the supervisory business model. He noted that ‘Supervisory technology is more than just technology; it’s a profound change in the business model and supporting activities.’

AI and ML are being deployed to augment various supervisory tasks, including compliance monitoring, risk assessment, fraud detection, and case management. Mariama Jalloh-Heyward, Programme Manager, Alliance for Innovative Regulation presented NextGenAI resources and highlighted that ‘AI-powered tools enable the extraction of insights from vast data sets, automating tasks like summarisation and trend identification to support data-led supervision. AI-driven solutions can not only predict but also prevent financial risks, redefining proactive consumer protection.’ For example, AI is being used to analyse cyber resilience policies, as discussed by Aristides Cavalcante, Head of Strategic Management and Specialized Supervision Department, Central Bank of Brazil, who outlined their usage of NLP (Natural Language Processing) to analyse cyber resilience policies from financial institutions to evaluate if the policies comply with regulations. Tools like AI-powered chatbots and virtual assistants are being explored to improve consumer access to information and enhance consumer protection. For instance, the Bangko Sentral ng Pilipinas launched an AI-powered chatbot called Bob in 2020 to streamline complaints handling. It can respond to complaints instantly and direct them to the relevant financial institutions. Bob’s data is used to analyse trends, identify consumer risks, and assist in proactive supervision of financial institutions

Dr. Sertaç Son, Head of Division, Prudential Domain Services, European Central Bank, provided caution against the hype around AI and the mislabeling of simple digitisation problems as AI initiatives. He advocated for prioritising high-value use cases, treating AI as an ongoing product rather than a one-time project, and adopting the 10-20-70 rule: 10% for algorithm development, 20% for enabling infrastructure, 70% for end-user adoption and integration into business processes. He also highlighted the importance of collaborative development between data scientists, engineers, and end-users to ensure meaningful and sustainable AI applications.

The sessions also showed a growing emphasis on the need for explainable AI (XAI) to ensure that supervisors who are not data scientists can effectively interpret model predictions. As outlined by Iota Nassr, Economist, OECD, “the lack of explainability in AI models makes it very challenging to mitigate risks, detect inappropriate activities, and maintain trust”. Marcus Lim, Assistant Managing Director, Banking and Insurance, Monetary Authority of Singapore highlighted that ‘The job of a supervisor does not change with AI. You’re still responsible for the safety and soundness of the financial institution, for protecting consumers. AI provides powerful tools, but supervisors must maintain the ability to question assumptions and challenge outputs.’

In summary, as technology becomes more pervasive, ethical considerations, bias mitigation and transparency become more vital for effective supervision. Babak Abbaszadeh, President & CEO, Toronto Center shared that ‘Technology is reshaping financial supervision, but AI is just one part of that. Supervisors need to ensure that the expansion of financial inclusion doesn’t lead to instability. AI is a powerful tool, but it must be used responsibly, with a strong ethical framework.’

Financial authorities should prioritise data-driven approaches to enhance supervision and risk management. This involves harnessing the power of data to move beyond traditional manual processes and embracing more objective and insightful methods of analysis and oversight. Sessions also highlighted the focus on using unstructured data such as text documents and call transcripts, and the potential of AI and Natural Language Processing (NLP) as crucial tools for extracting insights from these sources.

Generally, the focus lies on moving from manual reviews to objective data-driven approaches, using analytics to reveal patterns and insights that would otherwise go unnoticed. The Banco de España, for example, is viewing data as a core asset for supervisors, with a primary focus on data quality including monthly reconciliation of confidential reports and micro data.  Javier Tarancon, Head of the Information Analysis and SupTech Division outlined that ‘for us suptech is a tool to help supervisors get the most value out of very rich information sources available to them.’ In its latest SupTech Action Plan, the BdE  includes efforts to detect outliers and changing trends by using data to identify unusual patterns in financial data and monitoring for significant deviations, as well as to develop early warning indicators by using data to predict risks and improve the ability to anticipate problems.

The importance of high-quality, reliable data cannot be overstated. Data needs to be accurate, complete, and relevant to support supervisory tasks effectively. Financial authorities need to prioritise data quality and integrity as fundamental elements for effective suptech implementation. This involves putting in place the necessary systems, validation rules, and processes to ensure data is accurate, consistent, and reliable.
There is need for systems with built-in validation rules to preserve data integrity, ensuring that the data used for analysis is trustworthy and free from errors. Presentations in the session “Three Essential SupTech Solutions for Financial Consumer Protection” described how solutions defined complaint data points and validation rules using a “suptech data dictionary workbook”. Prester Mbiwa, Technology, Digital Finance & Data Analytics Lead, The SACCO Societies Regulatory Authority (SASRA) of Kenya noted that they are working on releasing a standard that can be used by ICT vendors to produce reliable reports.

This also includes having encrypted drives and user access controls to secure data, protecting sensitive information from unauthorised access and misuse and ensure data privacy. Michael Brand, who was involved in the development of FinTracer, access and systems controls in the session ‘Privacy-Preserving Algorithms for SupTech’. Feedback loops with data providers are also crucial to ensure ongoing quality and identify malicious links, enabling a proactive approach to data management that addresses issues as they arise. Moise Birigimana from the National Bank of Rwanda outlined their approach of sharing data back with the financial service providers during the data-cleaning process, which has led to improvements in data quality.

Financial authorities should actively work towards standardization and interoperability across different jurisdictions to foster scalability and reduce the cost of suptech development. This approach allows for greater efficiency, better data sharing, and ensures that solutions can be adapted and adopted more widely across the sector.

The need for interoperability was raised multiple times, including a call for alignment on standards and knowledge sharing, highlighting the importance of creating a unified approach that transcends organizational and geographical boundaries. Technology vendors are looking for standardization to enable scalability and reduce the cost of suptech development, which can foster innovation and more readily available solutions. The session on ‘Building Supervisory Powerhouses’ highlighted the European Single Access Point (ESAP) as a key example of a unified platform that aims to make financial and non-financial data accessible in an interoperable way.

Luay Gadallah, Chief Technology Officer, Eastnets noted the need for more research to enhance data interoperability standards across jurisdictions, and emphasized that developing global standards is crucial to ensure seamless integration across jurisdictions. Kevin Tsui, Manager, Hong Kong Monetary Authority agreed that ‘Sharing code and cloud architecture frameworks is a great way for regulators to quickly deploy and test solutions. Standardizing data protocols would enhance cross-border data sharing and improve suptech tools for supervision.’

The emphasis should be on delivering solutions that bring tangible impact quickly, enabling supervisors to respond more effectively and efficiently to emerging challenges. Iman van Lelyveld, Head of the Data Science Hub, Dutch Central Bank (De Nederlandsche Bank) stated that when balancing innovation and tangible impact, it’s important to start step by step by talking to each unit about key issues that need to be fixed and then “show with little examples how things could work”.

Arianna Brina, InsurTech Expert, Consumer Protection Department, EIOPA, stated that supervisors are encouraged to use a “bottom-up” approach, discussing problems with supervisory teams and addressing them with suptech.

• For instance, the Superintendencia Financiera de Colombia is using ‘Smart Supervision’ to move from manual complaint resolution to digital supervision, enabling continuous monitoring and quick analysis.
• The Reserve Bank of India developed an ML tool that identifies key risks deliberated in board and subcommittee meetings.
• Moise Birigimana, Manager, Financial Inclusion and Education, National Bank of Rwanda showcased a “high level Financial Inclusion Dashboard” they developed.
• Diana Zaig, Director, Advanced Analytics, Office of the Superintendent of Financial Institutions Canada presented a Temporal Fusion Transformer (TFT) for forecasting liquidity risks.
• Several speakers outlined suptech solutions deployed to address fraud detection and transaction monitoring challenges in digital payments.

With regards to leveraging suptech to address climate risks, Zooey Bossert, Data Scientist at the Dutch Central Bank, highlighted the application of “digital twin” technology to model and simulate physical climate risks within the financial system. Dr. Sean Carmody, Executive Director Policy and Advice Division, Australian Prudential Regulation Authority highlighted that climate risk requires a different approach to scenario analysis.

It is essential to recognise that suptech is not merely a technological challenge but a collaborative endeavor requiring input and expertise from various stakeholders and disciplines. A multi-faceted approach ensures that solutions are comprehensive, user-friendly, and meet the diverse needs of all parties involved, from technology vendors to policymakers and consumers.

As noted by multiple speakers, success of suptech approaches depends on collaboration and knowledge sharing across different geographies and roles, with an emphasis on creating a unified, multi-dimensional perspective. Maria Luciano, Centre for Digital Public Infrastructure, stressed the need for all perspectives to be considered in the development and deployment of suptech solutions and the importance of including “people being impacted” in discussions around online dispute resolution ecosystems. In the context of cyber risk supervision, Beju Shah, Head of the Nordic Centre, BIS Innovation Hub highlighted that “Public-private collaboration is essential to tackle the interconnected and borderless nature of cyber threats. Trust and reciprocity are the foundation of any successful cross-border collaboration in cybersecurity.”

Nydia Remolina, Assistant Professor of Law, Singapore Management University emphasised that this includes cross-departmental collaboration within institutions as ‘The success of suptech lies in fostering collaboration between multidisciplinary teams, including legal, technical, and supervisory experts, to navigate challenges like data management and AI implementation.’

It is essential to cultivate a culture of experimentation and collaboration to drive the development of innovative suptech solutions. By fostering an environment that is open to new ideas and partnerships, financial authorities can accelerate the development of impactful tools and strategies.

Experimentation and collaboration can influence senior leadership by providing tangible examples from other markets, showcasing the potential of innovative approaches. Financial Conduct Authority (FCA) in the UK has launched an AI Lab to understand the risks and opportunities around AI, demonstrating a proactive approach to exploring the frontier of regulatory technology. Fatima Abukar, Principal Advisor, Responsible AI and Data, noted that ‘the FCA is an innovative regulator, which has experimented with suptech through collaboration efforts like the digital sandbox, Tech Sprints, and the AI Lab. These initiatives help to see what is possible, what industry thinks, and how various organisations can mitigate risks.’ Similarly, the EU Digital Finance Academy was created to share knowledge, develop best practices, and implement strategies with EU national supervisors, exemplifying a commitment to collaborative learning and improvement. TechSprints and Hackathons were highlighted as effective methods for rapidly generating and testing innovative solutions.

Across sessions, speakers openly shared their challenges and learnings for others to build on. For example, Eugene Teves, Managing Director, Technology and Digital Innovation Office, Banko Sentral ng Pilipinas highlighted that “The biggest hurdle was understanding cloud contracts and adapting government procurement laws to accommodate cloud technologies, which had not been a priority until recently. Cloud allows us to scale our solutions more effectively, making it possible to deploy them to a larger number of institutions while reducing costs. The shift to cloud technology is just one step. New technologies will continue to evolve suptech solutions, and we must continue to embrace those changes and learn across sectors.”

Collaboration emerged as a pivotal theme, reinforcing the idea that progress in suptech cannot occur in isolation. Financial authorities should actively promote cross-border collaboration and knowledge sharing to accelerate the digital transformation of financial supervision globally. This includes engaging with other jurisdictions, sharing best practices, and collectively tackling shared challenges and opportunities.

Supervisors need to understand what other supervisors are doing around the world to learn from their experiences and avoid reinventing the wheel. Many speakers highlighted a desire to explore and adapt open-source technology, showcasing the collaborative spirit of the global supervisory community. Bernardita Piedrabuena, Vice President, La Comisión para el Mercado Financiero de Chile (CMF) highlighted the importance of data analysis and sharing among supervisors, especially concerning fintechs operating across borders.

The Bank of Spain demonstrates the value of collaborative action by sharing their tools with all their SSM (Single Supervisory Mechanism) colleagues, illustrating the benefits of a unified approach to suptech development and implementation. Alessandra Perrazzelli, Deputy Governor, Banca d’Italia also advocated for a shared monitoring scheme uniting regulators domestically and globally, to prevent fragmented approaches to AI governance and to avoid over-reliance on AI developers’ self-policing. She highlighted that “AI supports central banks by making sure that there is an enhancement in oversight of payment systems, improving information collection, and enabling real-time economic analysis. It is essential to foster coordination between authorities in designing a sound and responsible framework for AI adoption in the financial sector.”  Looking ahead she also emphasised that ‘’We should ensure that regulations are technology neutral, that are able to accommodate the fast pace of technological change without stifling innovation.”