BNM AI Feedback Statement Sets Out Responsible Adoption

BNM AI Feedback Statement Sets Out Responsible Adoption

Estimated reading time: 6 minutes · Last updated:

Bank Negara Malaysia has released a feedback statement on artificial intelligence in the country's financial sector. It sets out how the central bank intends to support responsible AI use after consulting 113 respondents. The BNM AI feedback statement answers a discussion paper the central bank issued on 5 August 2025, and it summarises what industry said about AI's opportunities, risks and governance. It ends with recommended considerations for financial institutions and names areas where firms can build capability together. BNM also singled out a forthcoming AI Model Risk Management Framework from the Asian Institute of Chartered Bankers as an example of the industry-led guidance it wants to see. No publication date for that framework has been given.

Key takeaways

  • 113 respondents: Bank Negara Malaysia's AI feedback statement draws on 113 submissions and wider stakeholder engagement. It responds to the central bank's discussion paper of 5 August 2025.
  • More than 70% adoption: Above 70% of Malaysia's financial service providers now operate one or more AI applications, according to figures BNM Governor Datuk Seri Abdul Rasheed Ghaffour gave in July.
  • Proportionate oversight: Respondents favoured responsible, proportionate AI oversight. They want it built on governance, accountability, data management and talent.
  • AICB framework pending: AICB's CROs Forum is preparing an ABM-endorsed AI Model Risk Management Framework. BNM said it will help institutions handle flawed or misused models.

What Bank Negara Malaysia's AI Feedback Statement Covers

A feedback statement is the document a regulator publishes after a consultation closes. It records what respondents told the regulator and explains what the regulator plans to do next. BNM's version closes a process that opened with its discussion paper on AI in the financial sector, dated 5 August 2025.

The central bank built its response on submissions from 113 respondents. It also drew on broader conversations with industry stakeholders that went beyond the formal replies. Those inputs covered three themes: where AI creates opportunity, where it creates risk, and what governance it demands.

The statement does more than summarise the feedback. It lays out BNM's own approach to backing AI innovation that is responsible. It flags areas where institutions can pool effort to build skills and swap good practice. It then finishes with considerations that BNM recommends financial institutions weigh as their AI use matures. BNM said it would keep working closely with industry as the technology advances, with the goal of strengthening trust and resilience, improving what consumers experience and helping the sector develop.

Where AI Is Already at Work in Malaysian Finance

BNM found that AI use is accelerating across Malaysian finance. Fraud management, risk assessment, customer service and operational efficiency are the areas where adoption is strongest. Fraud and risk work sit at the core of a lender's or insurer's business, so AI in those functions is no side experiment.

The scale is already significant. In July, Datuk Seri Abdul Rasheed Ghaffour, who governs BNM, put the share of Malaysian financial service providers using at least one AI application at above 70%. Because a single deployment is enough to count, that figure tracks how widely AI has spread rather than how deeply it is embedded.

The respondent pool reflected that spread. Insurers and takaful operators made submissions, as did payment service providers, technology providers, financial institutions and other industry participants. Takaful operators provide Islamic insurance structured to follow Shariah principles. Their presence means the feedback reached both conventional and Islamic finance.

Why Respondents Asked for a Proportionate Approach

Stakeholders broadly backed an approach to AI that is both responsible and proportionate. A proportionate regime generally scales the weight of oversight to the risk a particular use carries. Under that logic, a chatbot answering balance queries would not face the same scrutiny as a model that decides who gets credit.

Respondents named the foundations they consider essential. These were data management, accountability, governance, building talent and ongoing collaboration across the industry. Talent's place on that list matters. Governance rules mean little if institutions lack staff who can test, challenge and explain the models they run.

BNM's chosen tools fit that request. It frames its closing guidance as recommended considerations, and it does not describe them as new binding requirements. It also puts weight on firms working together. Whether that lighter framing holds as adoption deepens remains open. So far BNM has said only that it will keep engaging with industry as AI capabilities advance.

How AICB's Model Risk Framework Fits BNM's Approach

BNM welcomed industry-led efforts that translate emerging risks and practices into practical guidance. It named one example: the AI Model Risk Management Framework being prepared by the Asian Institute of Chartered Bankers (AICB). AICB's Chief Risk Officers' Forum, known as the CROs Forum, is leading the work. The Association of Banks in Malaysia (ABM) has endorsed it.

Model risk is the danger of losses or poor decisions that come from relying on a model. BNM said the new framework will help institutions manage AI model risks. Those include risks from models that are flawed and from models that are misused.

There is a precedent for this route. The CROs Forum, with support from BNM, earlier produced an AI Governance Framework for the Malaysian financial sector, and ABM endorsed that framework too. The pattern is consistent. The central bank sets direction, the banks' risk chiefs draft the operating detail, and the banking association endorses the result.

Malaysian AI guidance for the financial sector
Document Produced by Endorsement or support Status
AI discussion paper Bank Negara Malaysia Not applicable Issued 5 August 2025
AI feedback statement Bank Negara Malaysia 113 respondents consulted Published
AI Governance Framework AICB CROs Forum BNM support, ABM endorsed Previously developed
AI Model Risk Management Framework AICB CROs Forum ABM endorsed, welcomed by BNM Forthcoming, no date set

The case for and against BNM's approach

The case for

  • Adoption is already broad, with more than 70% of providers using at least one AI application, so guidance lands on existing practice rather than on theory.
  • The CROs Forum has delivered before: its AI Governance Framework, developed with BNM support, won ABM endorsement.
  • The coming model risk framework targets the specific dangers of flawed and misused models, which BNM itself highlighted.

The case against

  • The AICB model risk framework has no release date, so institutions lack dedicated model-risk guidance for now.
  • BNM frames its closing guidance as recommended considerations, so practice could vary widely between institutions.
  • Respondents listed talent development as essential, which points to a skills gap that frameworks alone cannot close.

What to be careful about

  • Until AICB publishes its AI Model Risk Management Framework, Malaysian institutions have no dedicated industry guidance on risks from flawed or misused AI models.
  • Because BNM frames its guidance as recommended considerations, institutions may apply it unevenly across fraud, risk assessment and customer service uses.
  • If institutions cannot develop the talent that respondents called essential, governance and accountability commitments may exist on paper only.

The bottom line

BNM's statement works less as a rulebook than as a map. It records that adoption is already broad, with more than 70% of providers running at least one AI application. It also shows that the 113 respondents want proportionate oversight anchored in governance, data and skills. The central bank's answer is to recommend rather than dictate, and to lean on industry bodies for the technical detail. That places the AICB model risk framework at the centre of what happens next, because it will deal directly with flawed and misused models. Until that framework is published, Malaysian institutions have principles and a governance framework. They do not yet have the dedicated model-risk guidance BNM has said it welcomes.

What to watch

  • Watch for the Asian Institute of Chartered Bankers to publish its AI Model Risk Management Framework; no release date has been set.
  • Watch whether BNM follows its pledge of continued engagement with further AI guidance for the financial sector; no timetable has been given.

Frequently asked questions

What is Bank Negara Malaysia's AI feedback statement?

It is BNM's response to its 5 August 2025 discussion paper on AI in the financial sector, built on input from 113 respondents. It summarises their views on AI's opportunities, risks and governance. It also closes with considerations BNM recommends financial institutions weigh.

Who is developing the AI Model Risk Management Framework in Malaysia?

The Chief Risk Officers' Forum of the Asian Institute of Chartered Bankers is driving it, and the Association of Banks in Malaysia has endorsed it. BNM welcomed the framework as an example of industry-led practical guidance.

Does Malaysia already have an AI governance framework for banks?

Yes. The AICB CROs Forum, with BNM's support, previously developed an AI Governance Framework for the Malaysian financial sector, and ABM endorsed it. The coming model risk framework adds detail on risks from flawed or misused models.



Share:

Categories

Newest course every month

Advertise your offline course to a wider audience with our landing page.

You May Also Like

BNM's AI feedback statement draws on 113 respondents, sets out its approach to responsible AI in finance and points to...
Cambridge's 2026 survey: 81% of financial firms use AI but only 14% call it transformational. Huawei's Fintelligent targets agent governance...
Koscom will present its tokenized securities platform, AI services including K-EFS, and MOU and test-bed results at Korea Fintech Week,...