RBI's Draft AI Rules Would Put a Board Member's Name on Every Model

RBI's draft model risk management guidance would apply to every AI/ML model a regulated entity uses — built in-house, bought from a vendor, or developed jointly. It calls for board-approved model risk frameworks, independent validation, continuous monitoring, stress testing, stronger cybersecurity controls, customer disclosure, and mandatory human oversight.

Shams Tabrej, co-founder and CEO of Ezeepay, expects the early cost to bite: "The cost of compliance is going to be high, especially in the early years." He says large banks, with governance already mature, will adapt faster than NBFCs and fintechs, who may need fresh investment in technology, people and independent oversight — some of which will show up in the price of services and vendor contracts.

Jitendra Tanwar, MD and CEO of Namdev Finvest, frames it differently: real upfront cost, yes, but an investment in institutional resilience rather than a tax. Srividhya Sridhar, group general counsel at Vivriti Next, points out that new-age NBFCs and fintechs already lean on automated decisioning more than legacy banks do — meaning the framework lands hardest on the firms most exposed to it, but also the ones with the most to gain from proving their governance early.

Tabrej expects a market response: growth in third-party AI governance software, model audits and bias testing, letting smaller institutions meet the bar without building compliance infrastructure from scratch.

Source: Business Standard