Name: Shivangi Kumari
Institute: Sardar Patel Subharti Institute of Law
Year: 3rd year (Vth Semester)
Email-Id: chaturvedishivangi80@gmail.com
ABSTRACT:
The rise of Artificial Intelligence (AI) in modern legal systems brings not only innovation but also deep confusion about fundamental concepts of criminal law. One of the most critical concerns is the idea of mens rea (guilty mind), it means that the mental element required to prove a crime. Artificial Intelligence is a machine which acts independently means without human emotions, intentions, or consciousness so how can AI be judged under a system built on human culpability? The rise of Artificial Intelligence raises questions about liability for crimes which an Artificial Intelligence commits, mainly because the Artificial Intelligence acts autonomously and with limited control humans or developers or programmers. This research paper explores the clash between AI’s functionality and the legal doctrine of mens rea, questioning whether the traditional framework of criminal liability can survive in the age of intelligent machines. The purpose of this paper is to evaluate legal, moral, and technical perspectives to propose a direction for future laws.
KEYWORDS:
Artificial Intelligence, Mens Rea, Criminal Liability, Legal Dilemma, Culpability, Autonomous liability.
INTRODUCTION:
Artificial Intelligence is no longer just science fiction; it’s deeply embedded in our everyday lives from self-driving cars to drones, from computer science to medical science and from artificially intelligent assistant on phones to artificially intelligent attorneys, and somewhere in legal system also, there is hardly any part of everyday life which remained untouched from the Artificial Intelligence. AI has helped to make human life easier, better, saving valuable time and energy. But with its growing presence or demand in our daily lives raises question for criminal law: Can a machine be guilty of a crime? If a machine that is AI found guilty of an act, then what kind of punishments be imposed on such an Artificial Intelligence entity. There are numerous legal issues which are yet to be settled. As AI continues to evolve, lawmakers and jurists around the world are struggling to address a legal system built for humans, not for machines. This research paper delves into the fundamental legal dilemma: Is mens rea still relevant now that machines are doing tasks once done only by humans?
RESEARCH METHODOLOGY:
This research methodology adopted here is a analytical and doctrinal, primarily based on the examination of existing laws, legal doctrine, case laws, and judicial pronouncements related to Artificial Intelligence and criminal liability.
I am elaborating the multi-dimensional approach I adopted in this paper:
- Doctrinal Research: The paper reviews primary sources such as statutes, constitutional provisions, and legal maxims related to mens rea and criminal liability.
- Comparative legal study: A comparative analysis has been carried out between jurisdictions like India, the United States, the European Union, and the United Kingdom, focusing on how these legal systems are beginning to conceptualize the role of AI in criminal justice.
- Descriptive and Analytical Approach: The paper describes the current legal status of mens rea and AI in criminal law, and further analyzes whether the existing principles are sufficient or require modification. It includes critical insights into why AI challenges the foundational theories of culpability.
- Interdisciplinary Perspective: Apart from legal texts, the study also incorporates perspectives from technology ethics, philosophy of criminal responsibility, and public policy to understand the real-world implications of AI induced harm.
- Secondary Sources and Expert Reports: Extensive use of reports by NITI Aayog, Law Commission of India, and the European Commission have been made. Scholarly articles, books, journals, and credible online resources are referred to for supporting arguments and building suggestions.
REVIEW OF LITERATURE:
Over the past few years, several scholars, jurists, and policy think tanks have begun exploring the intersection between AI and criminal law, some of the most insightful contributions are summarized below in easy way for the better understanding:
Gabriel Hallevy[1], in his work “when Robot Kill: Artificial Intelligence under Criminal Law”, presents three models of AI criminal liability and those are the perpetrator-via-another model, natural-probable-consequence model, and direct liability model. He argues that under certain frameworks, AI can be seen as fulfilling criminal intent indirectly.
Matthew U. Scherer,[2] in his article “Regulating Artificial Intelligence Systems: Risks, Challenges, Competencies, and Strategies”, emphasizes the regulatory vacuum around AI and highlights how the absence of legal structures could lead to misuse and misapplication, especially in criminal justice systems.
Ugo Pagallo,[3] a legal scholar in Italy, points out that existing criminal law is human-centric and cannot accommodate non-human agents like AI without significant reform. He suggests that instead of criminally punishing AI, accountability mechanisms should be placed on developers and users.
In Indian context, the NITI Aayog’s Report on Responsible AI (2021)[4] emphasizes ethical concerns and calls for a careful legislative framework that distinguishes between decision-making AI and human responsibility. However, it is silent on criminal culpability, reflecting a clear legislative gap.
A recent Harvard Law Review discussion (2022)[5] on “AI, Criminal Law, and the Future of Mens Rea” raised concerns that AI-generated decisions may bypass human moral reasoning, thus making it difficult to establish blame or intent under current legal doctrines.
CONCEPT OF MENS REA IN CRIMINAL LAW:
The foundation of criminal law rests on two Latin maxims” actus reus” (guilty act) and “mens rea” (guilty mind). For any individual to be held criminally liable, both elements must usually be present. The act alone is not enough; the law seeks to punish only those who act with a culpable state of mind such as intention, knowledge, negligence, or recklessness.
Mens rea ensures that individuals are held criminally responsible only when if they possess a blameworthy state of mind. For example, if someone accidentally steps on another’s foot, it may cause harm, but there’s no intent to hurt. On the other hand, if the same act is done intentionally, it reflects a guilty mind and therefore it attracts criminal liability. This mental element protects individuals from being punished unfairly. It allows courts to distinguish between genuine accidents and intentional crimes. That’s why even in serious offences like murder, proving mens rea (intention to kill or cause grievous harm) becomes crucial. However, the challenge with AI is that it lacks a mind, emotions, and moral reasoning. It can perform tasks, even make decisions, but it does not “intend” in the human sense. This raises a critical question: Can we apply human standards of mens rea to non-human entities like AI?
RISE OF ARTIFICIAL INTELLIGENCE IN CRIMINAL CONTEXTS:
AI is no longer limited to science fiction it is actively participating in real-world decisions. Some examples where AI plays a role in the legal or criminal field include:
- Predictive Policing: AI algorithms analyze crime patterns and suggest where future crimes might occur.
- Facial Recognition: Law enforcement uses AI to match faces in surveillance footage with criminal databases.
- Autonomous Vehicles: Self-driving cars have been involved in fatal accidents, raising questions of criminal liability.
- Military Drones: AI-driven drones can select and eliminate targets — sometimes with minimal human input.
- Social Media Monitoring: AI tracks hate speech, illegal activities, and even potential threats to national security.
These developments have immense benefits like- faster policing, reduced human error, and enhanced surveillance. But what happens when the algorithm makes a wrong call? What if a self-driving car kills a pedestrian because its sensors failed at that time? Can it be “punished”? Should the creator be liable for killing. In such scenarios, the traditional framework of mens rea seems insufficient. AI does not “intend” harm it merely follows data. Holding it liable like a human may seem absurd, but letting it escape responsibility altogether is equally dangerous.
CAN AI HAVE MENS REA?
There is a fundamental question: Can Artificial Intelligence ever possess a guilty mind? The answer is complicated, and most legal scholars would say no, not in the traditional sense. Unlike humans, AI systems operate on codes, commands, and machine learning algorithms created by the developers or programmers. They don’t have emotions, desires, or moral judgment. So how can they “intend” to commit a crime?
But here’s the twist some AI systems do learn, adapt, and even make decisions without human interference. For example, a self-driving car decides when to stop or accelerate based on real-time input, not just pre-programmed instructions. If that car hits a pedestrian, it raises legal and moral confusion which are given below:
- Is it a mechanical failure?
- Is it a design flaw?
- Or did the AI miscalculate in a way a human wouldn’t have?
Since mens rea is a human attribute, applying it to AI can present legal challenges and philosophical challenges. Some scholars suggest that instead of mens rea, we should talk about algorithmic foreseeability, whether the AI system could have reasonably predicted the harmful outcome. But even this is still a matter of debate.
Other frameworks include assigning a “corporate model of liability” to AI where the machine is not blamed directly, but the creators, users, or owners are held accountable, much like how companies are held liable for acts of their employees. So, while AI can never truly intend a crime, its decision- making capacity is forcing legal systems to consider the relevance and future of mens rea.
CHALLENGES IN CRIMINAL JURISPRUDENCE:
The entry of AI into the criminal space has shaken some of the oldest legal principles. Here are some key challenges mentioned that arise:
- Absence of Emotion or Consciousness:
Criminal law assumes that guilt and punishment go hand-in-hand with morality. But AI does not feel guilt or remorse, making traditional punishment (like imprisonment) meaningless.
- Attribution of Blame:
If an AI commits harm, who is to blame? The developer who coded it? The company that released it? Or the user? The chain of accountability becomes blurred or uncertain.
- Evidentiary Challenges:
AI systems are often “black boxes” even their creators can’t always explain how a particular decision was made. This makes it difficult to gather evidence or prove intention in court.
- No Moral Learning:
Human criminals may reform or express genuine remorse. While, AI cannot reflect on its actions or morally evolve unless reprogrammed. This challenges the rehabilitative function of criminal law.
- Existing Laws are Human-Centric:
Most criminal laws were written long before AI existed in our lives. They assume a human wrongdoer not Artificial Intelligence. Applying them to non-human entities like AI creates legal and philosophical confusion.
- Global Inconsistency:
There is no international consensus on how to handle AI in criminal law. While some countries are beginning to explore granting legal status to AI, others strictly reject the idea of granting legal status to AI.
These challenges show that the traditional criminal justice system especially its reliance on mens rea is not fully equipped to handle AI-related offences.
CASE STUDIES AND COMPARATIVE LEGAL INSIGHT:
To understand how different legal systems are responding to the AI-mens rea dilemma, it’s useful to look at a few real-life scenarios and international approaches. Here I am breaking down complex cases and concepts for better understanding:
1. The Uber Self-Driving Car Case (USA, 2018):[6]
In Arizona, a self-driving car operated by Uber hit and killed a pedestrian. The vehicle was in autonomous mode, and the safety driver was watching a video on her phone. Investigations found that the AI system recognized the pedestrian but didn’t classify her as a threat.
This is how the case raised major questions:
- Was it the fault of the AI?
- Should Uber be held liable?
- Or was the human backup driver negligent?
Ultimately, the driver was charged with negligent homicide, but no criminal charge was brought against Uber or the AI system. This case reveals the system’s current inability to handle autonomous liability.
2. Knight Capital Group Incident (USA, 2012):[7]
An AI trading algorithm went autonomous and caused $440 million in losses within 45 minutes due to a software bug. Though it wasn’t a criminal case, it highlighted how AI can cause massive harm without any malicious intent, simply due to operational decisions.
3. Germany’s Ethics Guidelines on Autonomous Cars (2017):[8]
Germany became the first country to issue ethical rules for autonomous vehicles, mandating that human life should always be prioritized over property or convenience. It also rejected assigning “legal status” to AI, stating that accountability must always lie with humans.
4. India’s Legal Silence:[9]
In India, AI-related legal developments are still in their infancy. While NITI Aayog has published reports on Responsible AI, there is no legislative or judicial framework for AI and criminal liability yet. Courts have not addressed mens rea in the AI context, and that is why it is leaving a significant gap.
5. European Parliament’s Proposal (2020):
The EU considered giving “electronic personality” status to advanced AI, which would allow assigning limited liability similar to corporations. This idea remains controversial but shows the global shift in legal thinking.
SUGGESTION:
Since the existing criminal law framework struggles to handle AI, here I am mentioning a few practical suggestions to move forward:
1. Establish AI-Specific Legal Categories:
Rather than trying to fit AI into human categories, lawmakers should create new legal definitions and responsibility models specifically for AI such as “automated agents” or “semi-autonomous actors”.
2. Developer and Operator Liability:
Accountability should lie with those who create, control, or deploy AI systems. This includes coders, manufacturers, and corporations. Just as a company can be held liable for a defective product, it should be liable for an AI that causes harm.
3. Algorithmic Transparency Laws:
Legal reforms should mandate that high-risk AI systems maintain decision-making logs so if harm occurs, it’s easier to trace why and how the decision was made.
4. Mandatory Testing and Ethical Audits:
Before AI is deployed in sensitive sectors like policing, military, or transport, it should pass ethical, legal, and safety audits, similar to drug trials or building inspections.
5. Reject “Personhood”, but Accept Accountability:
AI should not be treated as a legal person (like a human or company), but it should not be treated as a mere object either. A middle-ground model assigning indirect culpability is the need of the hour.
6. Judicial Training & Policy Update:
Judges, lawyers, and policymakers should be trained in tech-law interface, so that they understand how AI works and where its risks lie. Legal education must evolve AI related subjects for the law students.
CONCLUSION:
The increasing presence of Artificial Intelligence in decision-making, especially in areas affecting life, liberty, and safety, poses a fundamental challenge to the principles of criminal jurisprudence. Mens rea, once seen as the soul of criminal law, finds itself shaken by entities that act without thought, intention, or emotion. While AI can act, it cannot “intend” in a human sense. This disconnects human morality and machine functionality threatens the coherence of criminal law as we know it. The law cannot ignore the realities of AI, nor can it blindly apply human frameworks to non-human agents. A careful, balanced approach is essential one that preserves accountability, prevents injustice, and adapts legal doctrines to technological realities. The future demands a reimagined criminal legal system one that shifts focus from the offender’s mind to predictable harm, risk responsibility, and system design. Without such reform, we risk both technological misuse and legal irrelevance. AI may not kill mens rea entirely, but it certainly demands that we redefine what it means to be criminally responsible in a post-human world.
REFERENCES:
- Gabriel Hallevy, When Robots Kill: Artificial Intelligence under Criminal Law (2013).
- Matthew U. Scherer, Regulating Artificial Intelligence Systems: Risks, Challenges, Competencies, and Strategies, 29 Harv. J.L. & Tech. 353 (2016).
- Ugo Pagallo, The Laws of Robots: Crimes, Contracts, and Torts (Springer, 2013).
- NITI Aayog, Responsible AI: Strategy for India (2021), https://www.niti.gov.in.
- Proposal for a Regulation of the European Parliament and of the Council Laying Down Harmonised Rules on Artificial Intelligence (AI Act), COM/2021/206 final.
- German Federal Ministry of Transport and Digital Infrastructure, Ethics Commission Report on Automated and Connected Driving (2017).
- Harvard Law Review, AI, Criminal Law, and the Future of Mens Rea, 135 Harv. L. Rev. 1501 (2022).
- State of Arizona v. Rafaela Vasquez, CR2018-006085-002 (Maricopa Cnty. Sup. Ct. 2020).
- Knight Capital Group Technical Error Report (2012), U.S. Securities and Exchange Commission, https://www.sec.gov.
- Indian Penal Code, 1860 (Act No. 45 of 1860).
[1] Gabriel Hallevy, When Robots Kill: Artificial Intelligence under Criminal Law 12–14 (2013).
[2] 29 Harv. J.L. & Tech. 353, 360 (2016).
[3] The Laws of Robots: Crimes, Contracts, and Torts 88–91 (Springer, 2013).
[4] https://www.niti.gov.in/sites/default/files/2021-02/Responsible-AI-22022021.pdf.
[5] 135 Harv. L. Rev. 1501, 1504 (2022).
[6] State of Arizona v. Rafaela Vasquez, CR2018-006085-002 (Maricopa Cnty. Sup. Ct. Sept. 15, 2020).
[7] Knight Capital Group Technical Error Report, U.S. Sec. & Exch. Comm’n, https://www.sec.gov/litigation/admin/2013/34-70694.pdf.
[8] German Fed. Ministry of Transp. & Digital Infrastructure, Ethics Commission Report on Automated and Connected Driving (2017).
[9] Proposal for a Regulation of the European Parliament and of the Council Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act), COM/2021/206 final.