DSCI Is Building an AI Liability Framework for India

Avinash Mishra
By
Avinash Mishra
Business Correspondent
Avinash Mishra is a Business Correspondent at StartupFeed, covering quarterly earnings, banking and payments in India. He reports results from the country's largest listed companies alongside...
- Business Correspondent
DSCI says the draft will examine liability across users, integrators and AI providers, with a first version expected in September 2026.
Follow StartupFeed on Google News

The Data Security Council of India (DSCI) is building a framework to decide who is liable when an AI system causes harm. The question matters more as AI agents move from pilots into real business use. DSCI is the data protection industry body set up by Nasscom.

A first draft is expected in the third week of September 2026, DSCI chief executive Vinayak Godse said. It will cover three parts of what he calls the AI value chain.

The three parts are the organisations that use AI agents, the enterprises that build agents into their products, and the providers of AI tools and frameworks. Each layer touches a different point where something can go wrong.

“AI is a value chain, and distribution of liabilities is a great area of exploration,” Godse said. His point is simple. When AI is built by one firm, sold by a second and deployed by a third, no single party clearly owns the fault.

Godse said the liability question could also depend on the choices a company makes. That includes how a firm builds an AI system, whether it buys or builds the model, and how it deploys the agent. In his words, conditions matter in working out how liability is shared.

Liability clauses are already showing up in commercial contracts. Vishal Gondal, founder of fitness technology company GOQii, said most enterprise contracts now start with a blanket AI indemnity clause. A year ago, almost none did.

The strongest demand comes from banking, insurance, healthcare, government and public sector customers, Gondal said. These are the buyers with the most to lose if an AI decision goes wrong.

A company accepts responsibility for its own code, guardrails and an AI agent doing something. But the question of a wider AI failure is harder to place, Gondal said. He described it as a governance failure, not a product defect.

DSCI says the liability question is sharpest in financial services. There, AI is moving from helping staff to running systems that can start actions and transactions on their own.

What this means for you: If you build, resell or deploy AI agents in India, read your next enterprise contract’s indemnity clause closely, because the DSCI draft in September may reset who carries the risk.

Have a tip? Write to us at editorial@startupfeed.in.

Business Correspondent
Follow:
Avinash Mishra is a Business Correspondent at StartupFeed, covering quarterly earnings, banking and payments in India. He reports results from the country's largest listed companies alongside UPI and MDR economics, RBI regulation, and capital flows into spacetech, defence manufacturing and semiconductors. He joined StartupFeed's editorial team in 2026 and writes a regular markets brief for founders and operators tracking the public-market side of India's economy
Newsletter signup illustration: an open envelope with a letter and a paper plane

Don’t Miss Startup News That Matters

Join thousands of readers getting daily startup stories, funding alerts, and industry insights.

Newsletter Form

Free forever. No spam.