Federal Reserve vice chair for supervision Randal Quarles said the central bank is in the early stages of studying how the expanding use of machine learning in the financial sector may change its regulatory approach, but that so far it fits within the existing regulatory framework.
"We are paying a lot of attention ... within our existing framework. To the extent you have a machine learning tool that is interacting with customers we want to make sure that the traditional protections are being complied with," Quarles said.
While the Fed may need "more specialized responses" to the use of artificial intelligence or computer algorithms in assessing risk and making credit decisions, he said he was hesitant to impose "bounds" on what institutions can do, rather than develop "milestones" or circuit breakers that would pause any process that was producing bad results.