BSP drafting new rules to temper risks from AI

BSP Governor Eli Remolona

MANILA, Philippines – The Bangko Sentral ng Pilipinas (BSP) is preparing new regulations that aim to mitigate risks associated with the use of artificial intelligence (AI) in the local financial sector.

At a forum in Manila on Tuesday, BSP Senior Director Melchor Plabasan said the upcoming regulations would focus on ethical use of AI by banks and nonbanks, as well as on the “accuracy” of tools that are typically employed to help with decision-making.

READ: BSP launches cyber defense council

The new rules could be issued by the central bank in the first half of the year, he added.

“We have technology risk management; we have cybersecurity risk management; we have data privacy. So, all of these risks associated with AI have already been covered by existing [regulations],” Plabasan said.

“AI regulation that you can expect from the BSP is generally more clarificatory and at the same time, covering risk—the 20 percent of the risk that are not yet covered particularly on management of bias, accuracy and ethical use of AI,” he added.

According to the BSP official, “a lot” of local financial companies already have existing projects or are considering AI in their business road map.

Plabasan added that “more than one-third” of firms recently surveyed by the central bank have dedicated a “centralized” department for AI. These units are composed of experts like data scientists and data engineers, among others.

“AI is really exciting because it entails a lot of innovative ideas, but we should not lose sight of the risks associated with AI,” he said.

“What I can assure you is that the BSP is really committed to fostering an enabling regulatory environment that would promote responsible use of AI,” he added.

BSP Governor Eli Remolona Jr. earlier said that while AI adoption in the banking sector was “inevitable,” regulators and the industry must also understand the risks, including the tendency of generative AI to have “hallucinations.”

“They tend to imagine things that never happened. They are prone to herding. They tend to make the same recommendations to different institutions regardless of specific circumstances,” Remolona said in a previous speech. INQ