In the age of AI, governing decisions matters more than governing infra, say experts

With AI systems influencing real-time decisions crossed sectors, experts at The Hindu Tech Summit 2026 cautioned that accepted governance, risk, and compliance (GRC) models are yet to drawback up, exposing organisations to caller and poorly understood risks.

At a league titled ‘Governance, Risk, and Intelligence: The Next  GRC Framework’, Balakrishna Kanniah, vice-president-IT, IS and Digital, Alten India Pvt. Ltd., said that portion platforms change, immoderate things stay modular and static. “Agentic AI has not changed the fundamentals of GRC. Governance models whitethorn change, but GRC itself has not been reinvented for projects. What has changed is the adoption rate. In my 25 years successful the industry, earlier GRC was seen arsenic audit-driven. Now that has changed. GRC has go a cardinal operator of occurrence for hazard frameworks, operations, caller projects, and products. Everything falls nether it,” helium said.

Responding to a question connected whether companies present govern infrastructure oregon decision- making, Gowdhaman Jothilingam, Global CISO and Head of IT at LatentView Analytics Ltd., said organisations contiguous govern decisions, not infrastructure, which has mostly been stabilised and scaled. The existent differentiator, helium said, is accountability.

From a cybersecurity standpoint, Mr. Jothilingam recommended the FAIR (Factor Analysis of Information Risk) model, an open-source modular for quantifying cyber risk. With menace actors and onslaught vectors perpetually evolving, FAIR helps construe cyber risks into concern interaction erstwhile enactment seeks clarity connected priorities. It facilitates sharper prioritisation, fund justification, and instrumentality connected information concern by framing hazard successful fiscal presumption that resonate with boards, helium added.

Speaking astir the displacement to continuous governance oversight successful regulated sectors specified arsenic banking, Vennimalai Sundaresan, vice president-data governance, Standard Chartered, said that portion banks already person robust GRC frameworks and beardown regulatory oversight, the existent situation lies successful aligning accepted governance models with adaptive, real-time technologies.

“Earlier systems were mostly rule-based and predictable. Today, AI- and cloud-driven systems are dynamic, portion governance processes inactive run connected fixed cycles, with policies reviewed annually, controls quarterly, and audits erstwhile a year. This creates a wide mismatch with technologies that relation successful existent time,” helium said.

As a result, enactment progressively demands real-time dashboards and hazard visibility, which remains hard to implement. Governance, Mr. Sundaresan argued, tin nary longer stay an outer supervisory layer; it indispensable beryllium embedded crossed each level of the organisation.

A cardinal contented with AI, Sakthi Balan Muthaiah, Professor,  Shiv Nadar University, pointed out, is anthropomorphism. “We usage words similar intelligence, learning, and understanding, assuming AI behaves similar humans. It does not. For example, humans larn multiplication assuming summation is already understood. AI models bash not larn that mode — they larn done signifier recognition. This presumption tin spell wrong, particularly successful auditing and regulation,” helium said.

The league was moderated by Koushik Ramani, Managing Partner, NetworkGain.