By Enaya Agarwal
The law always knows who to blame, until now.
A self-driving car swerves to avoid one pedestrian and kills another. A machine malfunctions and misdiagnoses a patient. A chatbot gives financial advice that wipes out someone’s savings. In each case, we ask- Who is responsible?
Yet in today’s world, there may be no human responsible for this. The culprit is not a reckless driver, a negligent doctor or a fraudulent adviser, it’s an algorithm. Something, not someone. Our legal system was never designed to prosecute a computer code.
Modern law rests on a basic assumption: every action can ultimately be traced back to a human. Someone with intention, motive, negligence, accountability. Criminal law examines “mens rea”- the guilty mind behind an offence. Across legal systems, accountability is inextricably linked to personhood. The law asks not only what happened, but who made it happen.
This assumption is embedded in legislation itself. India’s Motor Vehicles Act, 1988, for example, states in Section 3 that “No person shall drive a motor vehicle in any public place unless he holds an effective driving licence…” The Act repeatedly refers to “the person driving the vehicle,” reflecting an era in which every vehicle necessarily had a human driver. Yet autonomous vehicles challenge this very premise. If a self-driving car, operating without human intervention, makes a fatal decision, who is “the person driving the vehicle”? The passenger? The manufacturer? The software developer? The company that trained the algorithm? The law, written for a human world, suddenly finds itself without a clear subject to regulate.
The problem extends far beyond self-driving cars.
Imagine walking into a hospital where an AI system misses the early signs of cancer. Or being denied a mortgage because an algorithm quietly labels you a financial risk. Perhaps you lose a job opportunity because an automated recruitment tool decides your résumé is less suitable than someone else’s. These scenarios are no longer science fiction, they are becoming part of everyday life. Artificial intelligence is already diagnosing diseases, screening job applicants, approving loans, drafting legal documents and, in some countries, even assisting judges with legal research and sentencing recommendations.
The difference is that these systems are no longer simply following instructions. They are making decisions that carry real legal, financial and human consequences.
And that presents the law with a question it has never had to answer before: what happens when the decision-maker isn’t a person?
Every legal system is built on the idea of accountability. We punish those who choose to break the law because they possess intent, judgment and moral agency. A driver who causes a fatal accident can be prosecuted. A doctor who acts negligently can lose their licence. A corporation that violates regulations can be fined millions.
But can you imprison an algorithm?
Can you prosecute a neural network?
Can software understand guilt, remorse or punishment?
Of course not.
An algorithm cannot stand trial, serve a prison sentence or pay damages from its own bank account. It cannot be deterred by the threat of legal sanctions, nor can it be morally condemned. Yet if its decisions cause devastating harm, society still expects justice. Someone must answer for the consequences but identifying that “someone” is becoming increasingly difficult.
Should liability fall on the software developer who wrote the code? The company that trained the model? The business that deployed it? Or the user who trusted its recommendations? In many cases, there is no obvious answer. Modern AI systems learn from enormous volumes of data and often operate as “black boxes,” producing outcomes that even their creators struggle to fully explain. The more autonomous these systems become, the more blurred the lines of accountability appear.
That uncertainty creates more than legal confusion, it creates opportunity.
History has shown that wherever the law is unclear, people find ways to exploit the gaps. Artificial intelligence is unlikely to be any different. Imagine a company whose hiring algorithm consistently discriminates against women or minority candidates. When challenged, the company shrugs: “We didn’t discriminate. The AI made the decision.” Or picture a criminal using AI-generated voice cloning to commit fraud before claiming the software acted unpredictably. As responsibility becomes increasingly dispersed between users, developers, software providers and corporations, the risk grows that AI will become less a tool for innovation and more a convenient shield against accountability.
The law cannot afford to let that happen.
Fortunately, governments are beginning to recognise the problem. The United Kingdom’s Automated Vehicles Act 2024 marks a significant departure from traditional road traffic laws by shifting responsibility, in certain authorised self-driving situations, away from the human user and towards an Authorised Self-Driving Entity responsible for the system. Rather than pretending a human driver still exists, the legislation acknowledges a new reality: if machines are making decisions, our legal frameworks must identify who remains accountable for them.
But legislation alone is not enough. As artificial intelligence becomes embedded in healthcare, finance, policing and transportation, regulation must evolve alongside it. High-risk AI systems should face stricter oversight than low-risk applications. Developers and companies must remain legally accountable for foreseeable harm. Transparency cannot be optional, especially when algorithms influence decisions that shape people’s lives.
Most importantly, governments must ensure that one phrase never becomes a valid legal defence:
“The AI did it.”
The history of law is one of adaptation. It evolved through the Industrial Revolution, the rise of corporations, the birth of the internet and the explosion of social media. Artificial intelligence presents the next great legal challenge, not because machines are becoming more intelligent, but because they are forcing us to rethink the very meaning of responsibility.
The question is no longer whether AI should be regulated. It is whether laws written for humans can continue to deliver justice in a world where some of the most consequential decisions are no longer made by humans at all.
I began with a simple question: Who is responsible? It is the question every courtroom asks and every victim deserves answered. Artificial intelligence should not change that. Our laws may have been written for humans, but our future demands that they evolve.
Because a world where no one is responsible is a world where justice cannot exist
