Agentic AI, Autonomous Decision-Making and Liability in India
Agentic AI refers to autonomous systems that can plan, make decisions, and execute complex multi-step workflows without constant human prompts. Agentic AI combines the capabilities of generative AI with additional features such as autonomous decision-making (ADM), memory systems, tool integration and adaptive learning techniques including reinforcement learning.
The deployment of agentic AI architectures across sectors in India from financial services, healthcare, e-commerce, telecoms, defence and other areas began accelerating during 2025 and 2026. Agentic AI is also rapidly transforming Indian manufacturing from medium- and small-scale industrial clusters in Punjab to massive automotive corridors in Tamil Nadu and Gujarat. Agentic AI assists with quality control, job scheduling, predictive maintenance, supply chain and logistics orchestration. A key feature of these deployments is that the AI system is no longer functioning merely as a conversational interface. Instead, it is interacting with enterprise software, triggering actions, coordinating workflows, making operational decisions and functioning with varying degrees of autonomy. This is why agentic AI is raising distinct legal and governance issues.
Despite the acceleration of multi-sectoral deployment of agentic AI systems in India in 2025-2026, Indian law does not yet contain a dedicated statutory framework governing autonomous decision-making (ADM) which is one of the core architectural characteristics of agentic AI. In the EU, Article 22 of the GDPR creates a specific right not to be subject to decisions based solely on automated processing where those decisions produce legal or similarly significant effects. The EU AI Act, in turn, is widely interpreted as embedding automated decision-making in its risk-based legal framework, but has no stand-alone category for ADM.
By contrast, India’s Digital Data Protection legal framework does not contain a provision similar to Article 22 of the GDPR. India has, at least for the present, not moved towards stand-alone AI legislation comparable to the EU AI Act.
Recent Indian policy documents engage with agentic AI only in a limited and largely generic manner. India’s AI Governance Guidelines released by the Ministry of Electronics and Information Technology in November 2025 define ‘Agentic AI’, but do not contain substantive guidance specifically relating to such systems. The Office of the Principal Scientific Adviser to the Government of India released a report on AI governance in January 2026 which limits itself to noting that agentic AI escalates privacy, security, safety and governance risks. A normative view is found in the Reserve Bank of India’s (RBI) Framework for Responsible & Ethical Enablement of AI (FREE-AI) Committee Report released on August 2025 which contains a recommendation to regulated entities that they put in place strong governance before deploying autonomous AI systems that are capable of acting independently in financial decision-making, including human oversight for medium and high-risk tasks. The RBI emphasizes that the entities that deploy the AI remain fully responsible regardless of the level of automation.
The absence of a dedicated statutory framework governing autonomous decision-making in India, or substantive policy guidance relating to it, is significant. It means that India, at least for now, is considerably more permissive toward automation and agentic systems than the EU. However, India may still regulate the consequence of autonomous decision-making indirectly through other legal doctrines such as negligence, consumer protection, sectoral regulations, cybersecurity duties or constitutional/public law principles in regulated sectors.
While European businesses often begin by asking whether an AI system is legally permitted to make a particular decision, in India businesses should ask who will bear responsibility if that decision causes harm. The EU approaches automated decision-making, at least in part, through the lens of individual rights. It remains to be seen how the obligations applicable to high-risk AI systems under the EU AI Act will ultimately apply to agentic AI. India has not adopted a comparable rights-based framework applicable to ADM. The legal debate may therefore develop less around whether automated decisions are permissible and more around where accountability sits when things go wrong.
The absence of a dedicated legal framework for automated decision-making does not necessarily give businesses greater freedom. It also creates uncertainty. Questions of liability allocation, standards of care and regulatory expectations may only become clear after an incident has occurred. In that environment, the way an agentic AI deployment is structured, governed and supervised becomes particularly important. Legal uncertainty does not reduce liability exposure. It increases the importance of upstream structuring.
As agentic AI systems become more autonomous, the central legal question may gradually shift from whether automation is permissible to who remains legally and operationally accountable once meaningful human supervision becomes increasingly nominal.
Aparna Viswanathan
AMALA AI Legal Consultants Srl
https://www.linkedin.com/pulse/agentic-ai-autonomous-decision-making-liability-india-viswanathan-4a8le