Who Is Responsible When An AI Agent Goes Wrong? India's DSCI Proposes Liability Framework
Last Updated: 2026-11-02 16:36:12
As companies increasingly deploy autonomous artificial intelligence agents to perform business tasks, a major question is emerging: who should be held responsible when an AI system makes a harmful decision or causes financial damage?
Show AI Summary
Summarized by AI.
DSCI proposed an AI liability framework for India.- Responsibility spans across the technology value chain.
- Autonomous AI risks require clear oversight and audit trails.
India’s Data Security Council of India (DSCI) has outlined an industry framework to address this question, proposing that responsibility should be assessed across the entire technology value chain rather than automatically assigned to a single organisation.
Latest and Popular Mobiles
The framework, titled “AI Liability: Industry Guidance for Agentic AI”, examines how accountability should be determined when autonomous AI systems operate across multiple platforms, technologies and business processes.
The development is particularly relevant as businesses move beyond conventional chatbots towards agentic AI systems capable of taking actions, interacting with external applications and completing tasks with limited human intervention.
Related Articles
Unlike a traditional chatbot that primarily generates responses to user prompts, an AI agent can be given a specific objective and carry out multiple steps to achieve it. Depending on the permissions provided, it may access information, use software tools, interact with other systems and make decisions during the process.
However, these capabilities also introduce new risks. An AI agent could potentially execute an incorrect transaction, expose sensitive information, provide an inappropriate recommendation or take an action that causes financial or operational damage.
Determining who should be held accountable in such situations can be complicated because multiple parties may be involved in building and operating the system.
These could include the developer of the underlying AI model, the company providing the software, organisations integrating the technology into their products and the business deploying the agent. Human operators and those responsible for configuring permissions may also play a role.
According to DSCI’s framework, assigning liability entirely to the organisation using an AI system or solely to an upstream technology provider could be legally flawed. Instead, responsibility should be examined based on the decisions made by different participants and the level of control they exercise over the technology.
The guidance also considers the different forms of impact that agentic AI systems could create and the factors that influence accountability across the deployment chain.
The framework does not introduce an entirely new set of legal standards. Instead, it explains how established legal principles can be applied to autonomous AI systems involving multiple vendors and interconnected components.
For Indian businesses, the guidance could become increasingly important as AI agents enter banking, healthcare, customer service, enterprise software and other sensitive operations.
Companies adopting agentic AI may need to pay greater attention to oversight mechanisms, access permissions, audit trails and clearly defined responsibilities. Simply deploying an AI system and allowing it to operate autonomously may not be sufficient when its decisions have real-world consequences.
The DSCI guidance provides an industry-focused approach to an issue that is likely to become more significant as AI systems gain greater independence in performing business tasks.