Blog

Written By: Apoorv Agarwal, Abhishek Kumar Jaiswal

Introduction

Artificial Intelligence (AI) has transformed sectors such as healthcare, banking, education, governance, and legal services. Its ability to analyse data and automate decision-making has improved efficiency, but it has also created new legal challenges. Since AI systems often function with minimal human intervention, existing laws are not always adequate to regulate their use. Therefore, legal systems must evolve to ensure accountability, transparency, and protection of individual rights.

However, Criminal Law works on a principal called “Actus non facit reum nisi mens sit rea”. That Latin maxim, which translates to “an act does not make a person guilty unless the mind is also guilty,” is the most foundational concept in criminal jurisprudence. 

Misuse of AI

The misuse of AI has become a serious concern. AI-generated deepfakes, voice cloning, and fake images are increasingly being used for fraud, cybercrime, identity theft, online harassment, and spreading misinformation. Such acts may attract liability under the Information Technology Act, 2000, particularly Sections 66C (identity theft), 66D (cheating by personation using computer resources), and other provisions relating to cyber offences. Depending on the facts, offences under the Bharatiya Nyaya Sanhita, 2023 (BNS), including cheating, forgery, criminal intimidation, and defamation, may also apply.

Privacy and Data Protection

AI systems depend on large volumes of personal data for training and decision-making. This raises concerns regarding consent, data security, and misuse of personal information. In India, the Digital Personal Data Protection Act, 2023 (DPDP Act) establishes a legal framework for the collection and processing of personal data based on lawful consent and accountability. The right to privacy has also been recognised as a fundamental right by the Supreme Court in Justice K.S. Puttaswamy (Retd.) v. Union of India (2017) 10 SCC 1, making privacy an essential consideration in AI regulation.

Intellectual Property Issues

Generative AI has raised important questions regarding copyright and ownership. AI models are often trained using copyrighted books, articles, photographs, music, and software without express permission from creators. This has led to debates over copyright infringement under the Copyright Act, 1957. Another unresolved issue is whether AI-generated content can receive copyright protection when there is little or no human authorship. These questions are currently being examined by courts across different jurisdictions.

AI in the Legal Profession

AI is increasingly being used for legal research, contract review, document analysis, and case management. While these tools improve efficiency, they cannot replace the professional Judgment of Advocates. Lawyers remain bound by the Advocates Act, 1961 and the Bar Council of India Rules, which require competence, diligence, confidentiality, and ethical conduct. Any AI-generated content must be independently verified before being relied upon in legal proceedings.

Burden of proof

Every facet of the case has some legal aspect which is pointed out by the court on which the parties have to prove or defend themselves. Similarly, every judicial proceeding ascertains some legal rights and liabilities which are interconnected with the facts and have to be proved before the court to satisfy them. Indian legislation provides such a statute with the title the Indian Evidence Act, 1872 under Section 101 to Section 111 which deals with the provision based on the ‘burden of proof’. The burden of proof is the legal burden or the obligation of the parties to prove the facts which further helps the court to decide in favor of either party. This burden of proof is also known as ‘Onus Probandi’. If the party on whom the burden lies fails to prove the burden then the case may go against him. There are Four types of Burdon of proof:

1. Beyond a Shadow of Doubt 
2. Beyond Reasonable Doubt 
3. On the Preponderance of Probabilities 
4. Near balance of Probabilities 

In “Dr. S.L. Goswami vs. State of Madhya Pradesh (1972) 3 SCC 22”, the Supreme Court of India cemented the “golden thread” principle of criminal law which in simple terms means that the burden of proving all ingredients of an offense rests entirely on the prosecution. The accused is entitled to the benefit of doubt if the prosecution fails to discharge this duty. This ruling relies heavily on Section 101 of the Indian Evidence Act, 1872, which mandates that the burden of proof always lies with the person asserting a fact. By adopting the foundational doctrine from “Woolmington v. Director of Public Prosecutions (1935)”, the Supreme Court affirmed that the presumption of innocence is paramount in criminal trials. 

Liability and Accountability

The principal types of criminal liability recognised in criminal law include:

1. Direct (Primary) Criminal Liability – Liability imposed on a person who personally commits the offence and satisfies the essential elements of the crime, including the required actus reus (guilty act) and mens rea (guilty mind).
2. Vicarious Criminal Liability – Liability imposed on one person for the criminal acts of another due to a legally recognized relationship, such as employer–employee or company–officer. In criminal law, vicarious liability is exceptional and arises only where expressly provided by statute (e.g., certain offences under corporate, environmental, or regulatory laws).
3. Joint or Common Liability – Liability of two or more persons who act together in the commission of an offence pursuant to a common intention or common object. Under Indian law, this is reflected in provisions such as common intention and unlawful assembly principles.
4. Abetment Liability – Liability arising from instigating, aiding, or intentionally facilitating the commission of an offence by another person, even if the abettor does not physically commit the offence.
5. Conspiracy Liability – Liability based on an agreement between two or more persons to commit an illegal act or a legal act by illegal means. The agreement itself constitutes the offence in many cases.
6. Attempt Liability – Liability imposed where a person, with the requisite criminal intention, performs an act towards the commission of an offence but fails to complete it.
7. Strict Liability – Liability imposed without the need to prove mens rea, where the statute specifically dispenses with the requirement of a guilty mind. This is generally confined to regulatory and public welfare offences.
8. Corporate Criminal Liability – Liability of companies and other juristic persons for offences committed through their directors, officers, or employees, subject to statutory provisions and judicial principles governing attribution of intent.

For a dissertation on AI criminal liability, the most relevant categories are:

1. Direct Criminal Liability;
2. Vicarious Criminal Liability;
3. Corporate Criminal Liability;
4. Strict Liability; and
5. Liability based on Abetment or Conspiracy (where AI is used as an instrument to commit offences).

A. International Approaches to AI Criminal Liability

International approaches to AI criminal liability differ considerably, as there is no uniform legal framework governing criminal responsibility for AI-related offences. Most jurisdictions continue to regard AI as a tool rather than an independent legal person and, therefore, impose criminal liability on the human actors responsible for its design, development, deployment, or use.

The European Union follows a strict liability approach in certain high-risk AI contexts, placing responsibility on developers, providers, or operators for harm caused by AI systems irrespective of proof of criminal intent. This approach prioritizes accountability and encourages responsible development and deployment of AI technologies.

In contrast, the United States adopts a causation-based approach, under which criminal liability arises only where a direct causal link can be established between the conduct of a human actor and the offence committed through the AI system. Accordingly, the prosecution must prove that the accused exercised sufficient control over the AI and possessed the requisite criminal intent under traditional principles of criminal law.

At the international level, regulatory initiatives increasingly emphasize transparency, explainability, fairness, data protection, cybersecurity, and ethical AI governance. International organizations, including the United Nations, the G7, and the OECD, have advocated common principles for responsible AI development. As AI-enabled crimes frequently transcend national boundaries, international cooperation through mutual legal assistance, information sharing, and harmonized legal standards has become essential for effective investigation and prosecution.

Thus, while the European Union emphasizes strict liability for those responsible for AI systems and the United States relies on a causation-based model grounded in traditional criminal law principles, the prevailing international position remains that AI itself cannot incur criminal liability. Instead, accountability continues to rest with the human individuals or entities responsible for its design, deployment, supervision, or misuse.

B. Proposed Regulatory Measures for Imposing Criminal Liability on AI in India

India does not presently have a dedicated legal framework governing criminal liability arising from the use of Artificial Intelligence (AI). Any future legislation should proceed on the principle that AI is not a legal person but a technological instrument, and that criminal responsibility must continue to rest upon the individuals or entities responsible for its design, development, deployment, operation, or misuse.

The proposed framework should adopt a causation-based model of liability, under which criminal responsibility arises only where a clear nexus is established between the conduct of the human actor and the offence committed through the AI system. Accordingly, the established principles of actus reus, mens rea, and causation should remain the foundation for determining criminal liability.

The legislation should further distinguish between offences that are “mala in se”, being acts that are inherently wrongful, and “mala prohibita”, being acts rendered punishable by statute or regulatory law. Such a distinction would enable courts to assess the degree of culpability and apply criminal sanctions in a manner consistent with settled principles of criminal jurisprudence.

In addition, the law should require reasonable standards of transparency, auditability, and human oversight of AI systems, while strengthening data protection and cybersecurity safeguards. A balanced statutory framework founded on these principles would ensure that technological innovation is encouraged without compromising accountability, public safety, or the fundamental principles of criminal law.

Need for a Comprehensive Legal Framework

India currently addresses AI-related issues through existing legislations such as the Information Technology Act, 2000, the Digital Personal Data Protection Act, 2023, the Bharatiya Nyaya Sanhita, 2023, the Copyright Act, 1957, and the Consumer Protection Act, 2019. However, these enactments were not specifically designed to regulate the unique challenges posed by Artificial Intelligence, particularly in the field of criminal liability. Therefore, India requires a comprehensive AI legislation that clearly defines criminal accountability based on a causation-based approach, prescribes obligations for developers, deployers, and users of AI systems, ensures transparency, human oversight, and auditability, and provides safeguards against algorithmic bias, cybercrime, and misuse of AI. Such a framework should balance technological innovation with public safety while remaining consistent with the fundamental principles of Indian criminal jurisprudence.

Conclusion

Artificial Intelligence presents both unprecedented opportunities and complex challenges for the criminal justice system. While AI is capable of performing autonomous functions, criminal liability should continue to rest upon the human actors who design, develop, deploy, or misuse such systems. India should enact a comprehensive AI law based on the principles of causation, accountability, transparency, and human oversight, while preserving the established doctrines of actus reus and mens rea. Such a framework would foster innovation without compromising public safety or the fundamental principles of criminal jurisprudence. As rightly observed, “What should keep us awake at night is not the fear that humans will become redundant, but the fact that humanity will.” This underscores the need to ensure that technological progress remains guided by human responsibility and the rule of law.