Bangladesh Bank set to introduce AI policy for banking sector by December
Cenbank plans to implement its own large language model AI
The Bangladesh Bank is preparing to introduce a comprehensive policy for the use of artificial intelligence (AI) in the country's banking sector by December, marking the first such initiative by a government institution.
A seven-member special working team has been formed to draft and implement the policy, which is expected to submit full guidelines next month.
The team leader, ICT Division-1 Executive Director Zakir Hasan, told TBS, "While AI enables significant benefits, uncontrolled use can lead to security risks and unethical practices. This policy aims to ensure that AI use in the banking sector and at Bangladesh Bank remains safe, ethical, and regulated."
Under the new framework, the central bank plans to implement its own large language model (LLM) AI system to prevent cross-border data transfer risks. "When data goes to foreign servers or networks, risks arise. Our in-house AI will mitigate these risks effectively," Zakir explained.
The policy aims to enhance economic forecasting capabilities, including predictions on inflation, GDP growth, and foreign currency reserves. AI will also serve as a tool to prevent fraud and laundering by analysing transaction patterns to detect unusual activity more quickly and accurately.
Additionally, AI will support monitoring of credit, market, and liquidity risks, enabling more data-driven and effective risk management. Integration of AI into fintech and digital banking initiatives is expected to improve customer-centric, safe, and efficient banking services.
Zakir further said AI can be used to present policies and reports in simple language for the public, and to analyse and resolve complaints more quickly.
Data from the Bangladesh Institute of Bank Management (BIBM) shows that 60% of banks currently lack an AI cybersecurity policy, while 40% have one in place. Moreover, AI is not incorporated in the operational policies of 68% of banks.
Research shows that 69% of banks are partially prepared, 11% are nearly ready, and 11% are fully prepared to implement AI. Conversely, 9% of banks are not prepared, and only 5% currently use AI in disaster recovery planning, leaving 95% without AI deployment in this area.
