Half of Kenyan banks adopt AI to curb credit risk and tackle cybercrimes, CBK survey shows
The survey shows that 83 per cent of institutions indicated that they were likely to adopt AI for credit risk assessment in future, and 82 per cent for cybersecurity, customer service, and e-KYC
A survey by the Central Bank of Kenya (CBK) has revealed 50 per cent of Kenyan banks have adopted Artificial Intelligence (AI) while all Credit Reference Bureaus (CRBs) indicated that they had not adopted the use of
AI in their operations. However, one out of three CRBs indicated that they were experimenting the use of AI and had allocated resources to support the adoption of AI in future.
The Survey on the Use of Artificial Intelligence in the Banking Sector also shows that 70 per cent of respondents indicated that they did not have AI Strategies, while 30 per cent had established AI Strategies.
The survey shows that 83 per cent of institutions indicated that they were likely to adopt AI for credit risk assessment in future, and 82 per cent for cybersecurity, customer service, and e-KYC.
“The leading risks reported by institutions that use AI include data quality, governance, and management (59 per cent), cybersecurity risks (54 per cent), limited numbers of AI-skilled staff (52 per cent), third-party dependencies (52 per cent) and AI governance risks (51 per cent),” the survey report says.
It also shows that 56 per cent of the institutions that had adopted AI had put in place measures to ensure the AI
models are explainable while 73 per cent of the institutions’ customers had equal access to the benefits of the AI models.
On the other hand, 78 per cent of institutions indicated that AI had enhanced their performance and operational
efficiency.
“Out of the institutions which had adopted AI, 41 per cent of the respondents had implemented AI policies while 59 per cent had not. 51 per cent of respondents indicated that they had established dedicated Data/AI teams,” it adds.
Out of the institutions that had implemented AI, 63 percent indicated that they actively monitored ethical considerations of their AI models to prevent unintended consequences for AI use while 37 per cent indicated they did not monitor these aspects.
At the same time, 68 per cent of survey respondents who had adopted AI indicated that they were using
generative AI Tools and frameworks in their operations.
The survey shows that 46 per cent of institutions developed AI applications in-house, 40 per cent outsourced
the development, and 24 percent partnered with other entities for development of AI applications.
It further indicates that 35 per cent of respondents used third-party service providers for AI/ML services while 17
percent indicated that the partnerships included AI and FinTech start-ups.
Based on Gartner’s AI Maturity Model Levels, 54 per cent of respondents indicated that they were at Level 1 of AI maturity (Awareness), with an early interest and few exploratory AI initiatives; 13 per cent of institutions were at Level 2 of AI maturity (Active), with pilots, experiments, and allocation of resources to AI while 19 per cent of institutions were at Level 3 of AI maturity (Operational), with some usage in workflows, and establishment of initial AI governance processes.
At the same time, 4 per cent were at Level 4 of AI maturity (Systemic), with use in workflows and operations for new business models while 1 per cent were at Level 5 of AI maturity (Transformational), leverage extensive use
of AI to create new products, services, and business models, achieving a competitive advantage.
According to Gartner’s AI Maturity Model Levels, Level 1 indicates awareness where organisations have an early interest in AI and are starting to have conversations around AI strategy, while are mostly experimental and exploratory.
Level 2 (Active) shows active initial experimentation and pilot projects with AI are underway where organisations begin to see the potential value of AI and start to allocate resources.
On the other hand, Level 3 (Operational) indicated that AI is used in at least one workflow or business process with organisations having established some AI governance and management practices.
Level 4 (Systemic) indicates that AI is present in the majority of workflows and operations and that AI initiatives inspire new digital business models and drive significant business value.
Level 5, which is considered a transformational stage, shows that AI is inherent in the DNA of the business organisations leveraging AI to create new products, services, and business models, achieving a competitive advantage.
According to the CBK survey, 9 per cent of the respondents, however, indicated that they had not considered using AI. The top three applications of AI and ML by respondents were credit risk assessment at 65 per cent, cybersecurity at 54 per cent, and customer service at 43 per cent. This was followed by electronic Know Your Customer (e-KYC) at 41 per cent and fraud risk management at 40 per cent.
The survey indicates that top three challenges in the adoption of AI were shortage of skills (46 per cent), high
costs and limited resources (46 per cent), and governance and regulatory compliance challenges (43 per cent).
It adds that 93 per cent of institutions surveyed recommended that CBK should issue Guidance on AI covering
governance and compliance, risk management frameworks and incident management and reporting.
The survey collected data on the state of innovation as of December 31, 2024, from 37 commercial banks, one mortgage finance institution, 14 microfinance banks, three CRBs and 70 Digital Credit Providers (DCPs).
According to the report, 92 per cent of commercial banks,100 per cent of mortgage finance banks, 100 per cent of CRBs and 91 percent of DCPs indicated that CBK should issue guidance on AI to support their AI developments.
The survey report illustrates that institutions have considerable interest in the adoption of AI and will continue to adopt AI to enhance operational efficiency and better serve customers.
It says that the adoption of AI has brought about operational efficiencies, particularly in credit risk management,
cybersecurity, and customer service.
Conversely, institutions face multiple risks as they adopt AI, including bias, inexplicability of models (black box),
and challenges in data and AI governance. A significant portion of the financial institutions reported the implementation of generative AI tools in their operations. This highlights the importance of ensuring that policy efforts cover the impact and risks of generative AI in the banking sector.
“As institutions adopt AI, it is necessary for CBK to provide guidance, particularly on best data
governance and AI governance strategies, AI risk mitigation measures, and interaction with third party AI vendors. Accordingly, CBK has embarked on the process of developing guidance on AI for the banking sector,” the report adds.
The survey respondents proposed that CBK provides guidance on ensuring data privacy, security, and ethical handling; model transparency, fairness in AI decision-making; addressing AI bias, discrimination, and
promoting ethical AI use; clarity on use of third-party AI models; managing third-party AI vendors; accountability structures for AI decisions; mitigating risks such as bias, unfairness, and lack of transparency; clear reporting lines for AI failures and regulatory reporting requirements.
