The Data Security Council of India has proposed an industry framework to help determine liability when autonomous AI agents cause harm, underscoring how quickly the legal and commercial risks around agentic artificial intelligence are moving ahead of regulation. The initiative comes as startups, enterprises and investors race to deploy AI systems that can take actions on behalf of users, from drafting code and processing transactions to making operational decisions with minimal supervision.
Liability Question
The central issue is no longer whether AI can generate content or recommendations, but who is responsible when an AI agent acts on flawed instructions, misinterprets context, or produces damage in the real world. That question is becoming urgent as autonomous systems are increasingly embedded in customer service, finance, software development, logistics and internal enterprise workflows. DSCI's proposal seeks to create a practical framework that can distinguish between the responsibilities of developers, deployers, platform operators and end users.
For India's startup ecosystem, the stakes are significant. Many young companies are building products around agentic AI because it promises lower operating costs, faster execution and new revenue models. But the same autonomy that makes these tools attractive also creates legal uncertainty. If an AI agent sends incorrect instructions, triggers an unauthorized payment, leaks sensitive data or causes a service outage, current liability rules may not clearly assign fault. That ambiguity can slow adoption, raise insurance costs and complicate enterprise procurement.
The framework is also likely to be watched closely by venture capital investors, who are backing a wave of AI startups but remain sensitive to regulatory and litigation risk. In the absence of a clear liability structure, investors may demand stronger contractual protections, more rigorous product testing and tighter governance controls before funding companies that rely heavily on autonomous systems. For larger enterprises, the issue is equally pressing: they need assurance that AI tools can be deployed without exposing them to open-ended legal exposure.
Why Framework Matters
DSCI's move reflects a broader policy challenge facing India and other major markets. Traditional technology law often assumes a human operator makes the final decision. Autonomous AI agents complicate that assumption by acting with varying degrees of independence, sometimes across multiple systems and with access to sensitive data or business-critical functions. A framework that maps liability across the AI lifecycle could help reduce uncertainty while preserving room for innovation.
Such a framework would likely need to address several practical questions: whether liability should depend on the level of autonomy granted to the system, whether developers should be accountable for foreseeable failures, and how much responsibility should rest with the organization that deploys the agent in a production environment. It may also need to consider whether users who give broad instructions, or fail to supervise the system adequately, should share responsibility when harm occurs.
The proposal arrives at a moment when global regulators are increasingly focused on AI governance, safety and accountability. India has so far taken a relatively measured approach, favoring industry consultation and risk-based discussion over sweeping restrictions. DSCI's framework fits that pattern, offering a possible template for self-regulation or soft-law guidance before formal legislation catches up.
Startup Stakes Rise
For startups, the framework could become a valuable operating guide if it is adopted widely by industry and referenced by customers, insurers or courts. Clearer liability rules would help founders design products with better safeguards, document decision trails and define contractual boundaries with clients. It could also encourage more responsible deployment by making it easier to identify where human oversight is required and where automation can safely operate.
At the same time, any framework that is too vague may fail to resolve the core problem. If liability is spread too broadly, it could discourage innovation. If it is too narrow, it may leave harmed users without meaningful recourse. The challenge for DSCI will be to strike a balance that supports experimentation while ensuring accountability when autonomous systems fail.
The proposal is an early but important signal that India's technology ecosystem is entering a new phase of AI governance. As AI agents become more capable and more deeply integrated into business operations, the question is shifting from what these systems can do to who pays when they go wrong. DSCI's framework suggests that the market can no longer postpone that debate.