Artificial intelligence is no longer a futuristic concept confined to science fiction. It is increasingly shaping decisions that affect our daily lives. From determining creditworthiness and screening job applications to assisting in medical diagnoses and generating information consumed by millions. As AI systems become more sophisticated and autonomous, an uncomfortable legal question is emerging: who is responsible when an AI system causes harm?
The question is no longer theoretical. AI tools have generated defamatory content, produced dangerous misinformation, exhibited bias in decision-making, and, in some cases, influenced critical decisions with real-world consequences. As AI becomes more deeply integrated into society, instances of harm are likely to increase. Yet our legal systems remain ill-equipped to answer a fundamental question: when an AI system “goes rogue,” who should be held accountable?
At the outset, one principle must be clear. AI itself cannot be held legally responsible. Unlike human beings or corporations, AI systems have no legal personality. They cannot be sued, fined, imprisoned, or ordered to compensate victims. While AI may appear capable of independent reasoning and decision-making, it remains a tool created, trained, deployed, and controlled by humansor organisations. The real question, therefore, is not whether AI is responsible, but which human actors or organizations should bear responsibility for the harm it causes.
This challenge is compounded by what many describe as the “accountability gap.” Modern AI systems often operate as black boxes, producing outputs through complex processes that may not be fully understood by the users. When an AI-generated recommendation causes harm or an automated decision results in discrimination, identifying the precise source of the problem can be difficult. Was the fault in the design of the model? The training data? The deployment environment? Or the manner in which the user employed the system? The law cannot permit technological complexity to become a shield against accountability. If harm occurs, responsibility must still be traceable to a human decision-maker or corporate entity.
In reality, liability is unlikely to rest on a single actor. A more practical approach is to view AI accountability as a shared responsibility. Developers who design and train AI systems should be responsible for negligent design, inadequate testing, foreseeable risks, and failures to implement appropriate safeguards. Companies that deploy AI systems should be accountable for ensuring that these tools are used responsibly, monitored effectively, and accompanied by adequate protections against misuse. At the same time, users cannot be absolved of responsibility merely because an AI tool was involved. A person who knowingly employs AI to commit fraud, spread misinformation, create deepfakes, or engage in other unlawful conduct should remain fully liable for those actions.
The common thread linking these actors is control. Those who design, deploy, profit from, or misuse AI systems exercise varying degrees of control over the technology. Liability should correspond to that control. This principle is neither radical nor unprecedented. Product manufacturers, employers, and service providers are routinely held accountable when their actions or omissions cause harm. AI should not become an exception simply because it is technologically advanced.
A crucial aspect of this debate is transparency. Meaningful accountability is impossible without understanding how an AI system arrived at a particular output. One of the most significant shortcomings of many AI systems today is their opacity. When harmful or inaccurate content is generated, users often have little visibility into the sources, data, or reasoning processes that produced the result.
Greater transparency in AI design and outputs is therefore essential. Wherever feasible, users should be able to identify the sources relied upon by an AI system in generating a response. Equally important is the ability to trace how information moved through the system before producing the final output. Such traceability would not only help users assess the reliability of AI-generated information but would also provide courts and regulators with valuable evidence when disputes arise.
If an AI system causes reputational, financial, or physical harm, a transparent record could help determine whether the problem originated from flawed training data, inaccurate source material, defective system design, inadequate safeguards, or deliberate misuse by the user. In the absence of such transparency, assigning responsibility becomes significantly more difficult. Accountability requires explainability. Without it, victims may find themselves unable to identify who is responsible, while companies may seek refuge behind the complexity of their algorithms.
India’s existing legal framework is not fully prepared for these challenges. Principles of contract law, consumer protection, tort law, and information technology regulation may provide partial remedies, but they were not designed for autonomous AI systems capable of generating unpredictable outcomes. As AI adoption accelerates across sectors such as healthcare, finance, transportation, and public administration, the gaps in the current framework will become increasingly apparent.
India does not necessarily need a comprehensive AI liability statute overnight. However, it does need a clear regulatory framework that assigns responsibility across the AI ecosystem and imposes higher standards on high-risk applications. Such a framework should prioritise accountability, transparency, and traceability while ensuring that innovation is not stifled.
Artificial intelligence may be capable of making decisions, but it cannot bear legal or moral responsibility for them. As AI systems become more powerful, the law must ensure that accountability remains firmly human. Otherwise, we risk creating a future in which harm is real, victims are identifiable, but responsibility disappears into the black box of artificial intelligence.
About The Author
Harsh Singh Dahiya is an Advocate at the Supreme Court of India and a legal and political commentator who appears regularly on leading news platforms including CNN-News18, Times Now, NDTV and NewsX.
His work focuses on Indian politics, national security and international affairs, with particular attention to questions of sovereignty, state power and India’s position in the evolving global order.