Already have Rappler+? Sign in to listen to groundbreaking journalism.

This is AI generated summarization, which may have errors. For context, always refer to the full article.

In a 2024 BSP survey, 21 out of 48 responding financial institutions, or 44%, said they had already deployed at least one AI system, while 60% included AI or machine learning in their technology roadmaps.

MANILA, Philippines – The Bangko Sentral ng Pilipinas (BSP) has issued voluntary guidelines on how banks, e-wallet operators, and other supervised financial institutions should use artificial intelligence.

The guidelines, issued through Memorandum No. M-2026-031, lay down what the BSP calls its "minimum supervisory expectations" for the responsible use of AI. However, they are non-binding, meaning financial institutions are encouraged — but not expressly required — to adopt them.

AI is already a part of Filipinos' everyday financial transactions, even when customers don't immediately notice it.

Financial institutions can use AI to analyze payment and spending patterns, estimate whether a borrower is likely to repay a loan, verify a customer's identity during an online application, recommend financial products, detect possible scams, and respond to concerns through chatbots.

In a 2024 BSP survey, 21 out of 48 responding financial institutions, or 44%, said they had already deployed at least one AI system, while 60% included AI or machine learning in their technology roadmaps.

The most common uses included fraud and anti-money laundering monitoring, electronic know-your-customer checks, credit-risk scoring, personalized product recommendations, and generative AI tools. This was two years ago; this usage rate is likely to only have gone up.

Some institutions have also publicly disclosed how they use the technology. GCash says AI and machine learning may help it detect scams, fraud, money laundering risks, and unauthorized access. The e-wallet may also analyze transaction patterns and other signals to determine whether certain services are suitable or available to a user.

UnionBank, meanwhile, says it uses AI in customer-facing chatbots and to assist employees making decisions that may affect an account. It also says it may use AI and machine learning for real-time fraud detection. (READ: Beyond OTP: UnionBank uses voice authentication, AI to fight fraud )

The growing use of algorithms is why the BSP introduced five governance principles, summarized by the acronym STARS: sustainability, transparency, accountability, responsibility, and security.

"AI is spreading across BSFI operations and STARS provides them with principles that can help them innovate while mitigating unintended consequences from the use of the technology," said BSP Deputy Governor Lyn Javier in a press release on Tuesday, July 14.

"We want BSFIs to take advantage of AI, especially to serve their customers, and do so while being guided by developing global standards."

Among the most important principles is that customers should be notified when AI output is being used in a product or process. Institutions should also disclose relevant limitations, particularly when an AI result could be misinterpreted or misused.

The BSP said users should be able to question how an output was produced and why a recommendation was made, instead of blindly relying on the technology.

Banks and e-wallets must also maintain human oversight. While AI can produce predictions or recommendations, the memorandum says humans remain ultimately accountable for the resulting decisions.

"While AI systems provide recommendations, humans are ultimately accountable for decisions made. The output of AI systems should not replace or diminish human responsibility," the BSP said in its memorandum.

The BSP guidelines doesn't explicitly stop financial institutions from using AI in credit decisions. An algorithm may analyze a borrower's income, transactions, payment behavior, or other available data and recommend whether an application should be approved or rejected.

In practice, this means AI may indeed play a big role in turning down a loan application. But the BSP is clear that financial institutions can't simply leave the final responsibility to the machine. Human oversight must remain in place, and a person or team must ultimately be accountable for decisions made with the help of AI.

Banks must also establish safeguards, assign clear ownership of AI systems, and monitor their results for errors and unfair outcomes.

"AI systems may perpetuate biases, leading to unfair practices and the exclusion of individuals from access to financial products and services," the BSP acknowledged in its memorandum.

This risk may arise when an AI model is trained using incomplete, inaccurate, or unrepresentative information. Such a system could disadvantage certain groups, including low-income borrowers, minorities, vulnerable consumers, or people with limited access to formal financial services.

To prevent this, the BSP recommended that training data be "well-prepared, well-represented, and free from unauthorized interference," while ensuring that AI systems do not cause harmful effects on any demographic, particularly minority or vulnerable groups.

Institutions must likewise protect personal data, provide clear opt-in or opt-out mechanisms where applicable, and monitor systems for hallucinations, or instances when generative AI produces false or invented information.

The BSP opted for voluntary, principles-based guidance because institutions differ widely in size, technological capability, and how extensively they use AI.

At the same time, perhaps imposing too strict a framework could stifle innovative and experimental use of AI in the industry, which is still at an early stage.

Currently, there's no specific compliance deadline or penalty for failing to adopt a formal AI governance framework.

The BSP also would "monitor the developments in the field and, whenever necessary, issue appropriate regulations or policies to foster innovation and preserve the stability and competitiveness of the financial system," leaving open the possibility that the current voluntary guidelines could eventually form the basis of stricter, enforceable rules. – Rappler.com

Share in Chat Share article Facebook X (Twitter) Copy Link Copied How does this make you feel.