Setty said the growing use of agentic AI in financial services could eventually require institutions to think about “Know Your Agent”, with clear mechanisms around an AI agent’s identity, authentication, consent, transaction limits, audit trails and revocation.
“Banks have spent decades building robust processes about know your customer. As agents begin to participate in financial transactions, we will increasingly need to think about ,” Setty said at the .
Also read:
From recommendations to execution
Agentic AI represents the next stage of artificial intelligence, where systems can move beyond assisting with tasks or carrying out a defined set of instructions to acting with a degree of autonomy, Setty said.
This shift makes accountability more critical, particularly when AI moves from making recommendations to executing actions, he added.
There must be an audit trail, traceability and the ability to understand why an important action was taken, according to the SBI chairman.
The banking sector has already deployed AI and machine learning across areas including credit assessment, cash flow-based lending, early warning systems, fraud detection, anti-money laundering monitoring and customer service. Agentic AI could extend these capabilities across the financial lifecycle, including KYC, loan appraisal, underwriting, reconciliation, complaint management and customer servicing.
‘AI versus AI’ risk
The growing deployment of autonomous systems could also create an “AI versus AI” environment, Setty said, requiring banks to combine identity intelligence, behavioural intelligence, transaction intelligence and real-time risk assessment.
The risks could be amplified because errors made by autonomous systems can trigger a sequence of actions across multiple connected systems at machine speed, potentially affecting customers, counterparties and institutions, he said.
“The same capability that gives AI its enormous power can therefore amplify the consequences of a mistake,” Setty said.
Trust cannot be an afterthought
Setty said trust must be embedded into agentic AI models rather than added later. He outlined what he called the three A’s for deploying AI in financial services: accuracy, accountability and access without asymmetry.
An AI agent handling financial services would need to be accurate consistently and at population scale, while systems must also ensure accountability for their actions, he said.
AI should also be accessible without bias, with technology adapting to customers rather than requiring customers to adapt to technology.
India’s next challenge: building trust at scale
India’s digital public infrastructure has demonstrated that technology can be adopted rapidly when it is simple, affordable, interoperable and trustworthy, Setty said.
As agentic AI enters banking, however, no single institution will be able to build the required trust and scale alone, he said, calling for a broader coalition involving regulators, banks, fintechs and technology companies.
India’s next opportunity, Setty said, is to move from digital inclusion to “intelligent inclusion” by making intelligent financial services available at population scale.
as a Reliable and Trusted News Source
(You can now subscribe to our )