The Data Security Council of India (DSCI) has released an India-focused AI Liability Industry Guide for Agentic AI. It tackles a key question. When AI agents act on their own and something goes wrong, who bears the consequences?
The guide gives a practical view of how liability, accountability and responsibility should be understood as enterprises adopt agentic AI. It also looks at how they may be allocated. DSCI intends it as a reference for the safe use of AI inside enterprises.
It is a live discussion paper. DSCI will update it from time to time, based on input from industry sectors.
Existing Rules, Not New Standards
The guide does not propose new standards. It explains how established binding doctrines apply to autonomous AI systems that involve several vendors. It also maps where responsibility sits across the value chain that brings AI into use.
It covers the domains of agentic AI, the forms of impact and risk these systems can create, and the factors that shape accountability. It argues that a proportionate, clearly allocated approach to governance, oversight and liability matters for India. Such an approach would guide user industry while speeding up responsible AI adoption.
DSCI is also asking for feedback on how this allocated approach spreads across governance, oversight and liability in India.
A Call to Move From Principle to Practice
Vinayak Godse, CEO of DSCI, said the guide tries to move AI governance and safety conversations from principle to practice. He made three points.
First, the discussion must reach the ground level. It should include every player in the AI value chain, from those who build models to those who deploy and use them. Second, intent alone is not enough. Organisations need clear decision spaces and tools they can actually put into use. Without these across the value chain, liability talks will not lead to practical, sector-specific action. Third, AI brings conditions and risks that no one can yet foresee. The framework must stay adaptive enough to handle them as they emerge.
Godse said DSCI looks forward to working with the government and industry to turn the guidance into action.
Who Should Respond
The consultation matters most to end-user organisations, service providers and integrators, product companies and start-ups, tool and framework providers, and compute and intelligence providers.
DSCI encourages industry leaders to test the guide against real deployments. That includes the C-suite, AI officers, government bodies, research think tanks, academic institutions, legal bodies and chief risk officers. It wants them to say what strains, what is missing and what does not hold.
