LONDON (Reuters) - Banks must anticipate risks from using artificial intelligence (AI) and machine learning (ML) in their operations as part of their day-to-day governance, a top global banking regulator said on Wednesday.

There are unanswered questions on whether the use of AI or ML in banking is a net positive or negative to global financial stability, said Bank of Spain Governor Pablo Hernandez de Cos, who chairs the international Basel Committee on Banking Supervision.

"My main message is that the use of AI in banking raises important prudential and financial stability challenges," de Cos said in a speech in Washington.

"Left unchecked, such models could potentially amplify future banking crises."

Digital innovation will further fuel cross-border and cross-sectoral financial interconnections, requiring collaboration among central banks and regulators to achieve an appropriate regulatory baseline to oversee the use of AI and ML, de Cos said.

"When it comes to banking, it is critical that banks anticipate and oversee the risks and challenges posed by AI/ML - both at the micro and the macro level - and incorporate them in their day-to-day risk management and governance arrangements," de Cos said.

The Basel Committee will soon publish a more comprehensive report on the digitalisation of finance and its implications for regulation and supervision, he said.