Banks will need to develop a “know your agent” framework covering identity, authentication, customer consent, transaction limits and audit trails as artificial intelligence begins to participate directly in financial transactions, State Bank of India Chairman CS Setty said on Thursday.
“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 know your agent,” Setty said at the Global Fintech Fest 2026.
ALSO READ
Such a framework would require mechanisms for identity, authentication, consent, transaction limits, audit trails and revocation, he said. Banks may also have to prepare for an “AI versus AI” environment by combining identity, behavioural and transaction intelligence with real-time risk assessment.
Setty said AI agents could be deployed across the financial lifecycle, including fraud detection, KYC and anti-money laundering processes, loan appraisal, underwriting and reconciliation. On the customer side, AI could improve service, complaint management and customer lifecycle management.
STORIES YOU MAY LIKE
However, greater autonomy also increases the potential impact of errors. “When a system can act autonomously, an error can potentially trigger a sequence of actions across multiple systems and affect customers, counterparties, and institutions at machine speed,” he said.
Setty said trust in agentic AI would have to be built around three principles — accuracy, accountability and access without asymmetry. In financial services, AI systems cannot operate with merely average levels of accuracy, he said, adding that human oversight would remain necessary for complex and high-risk financial decisions even as routine, high-volume and data-intensive activities become increasingly automated.
He also flagged the cost of computing as a key challenge to deploying generative and agentic AI at scale in India. While the initial fixed cost of AI could be high, low incremental costs could make large-scale deployment viable. India would, however, need affordable AI models and computing infrastructure.
ALSO READ
“How do we design AI models and AI computing power for India, which is affordable, is going to be an important element when we talk about scale to be achieved here,” Setty said.
For the next phase of financial inclusion, Setty highlighted voice-based conversational banking, particularly in Indian languages, as a potential use case for customers who are more comfortable speaking than typing.
At SBI, AI is already being used in customer service, pre-approved personal loans, credit assessment, cash-flow-based lending, fraud detection, anti-money laundering monitoring and early-warning systems, he said.
Read Next
This article was first uploaded on September ten, twenty twenty-six, at twenty-nine minutes past five in the evening.
© IE Online Media Services (P) Ltd
aniltikotekar4sme.com 