Author: Arnav Gupta
College: Bharati Vidyapeeth’s Institute of Management and Research
LinkedIn: https://www.linkedin.com/in/arnav-gupta-8377771b6
TO THE POINT
Imagine a courtroom where a person’s liberty depends upon a video recording. The facial expressions appear natural, the voice sounds authentic, and every detail seems convincing. Yet the video was not recorded by a witness, captured by a camera, or created by a human being. It was generated entirely by artificial intelligence.
Only a few years ago, such a scenario would have belonged to science fiction. Today, it represents a growing legal reality. Artificial intelligence is rapidly transforming the way information is created, processed, and presented. From AI-generated transcripts and predictive analytics to image enhancement technologies and synthetic media, artificial intelligence is increasingly influencing the evidence that may eventually reach courtrooms. As these technologies become more sophisticated, legal systems across the world are being forced to confront a difficult question: can evidence generated or modified by artificial intelligence be trusted?
The administration of justice has always depended upon evidence. Courts determine facts, assign liability, and resolve disputes based upon material that is capable of establishing the truth. Whether in criminal trials, civil disputes, constitutional litigation, or commercial arbitration, the credibility of judicial outcomes depends largely upon the reliability of the evidence placed before the court. Consequently, legal systems have developed detailed rules governing admissibility, authenticity, relevance, and evidentiary weight. These safeguards exist because justice requires more than information; it requires trustworthy information.
Artificial intelligence challenges many of these traditional assumptions. Unlike conventional evidence, AI-generated outputs may emerge from systems that operate through complex algorithms and machine-learning processes that are not always transparent or easily understood. In some instances, even the developers responsible for designing the technology may be unable to explain precisely how a particular output was generated. This phenomenon, commonly referred to as the “black box problem,” creates significant difficulties for courts seeking to evaluate the reliability of AI-generated material.
The issue becomes even more complex when one considers the growing capabilities of generative artificial intelligence. Modern AI systems can create realistic photographs, fabricate convincing audio recordings, generate human-like text, and produce videos that are increasingly difficult to distinguish from genuine recordings. The rise of deepfake technology has demonstrated that artificial intelligence possesses the capacity to create highly persuasive but entirely fabricated evidence. In a legal system where truth and authenticity are paramount, this development presents an unprecedented challenge.
At the same time, artificial intelligence also offers significant benefits. AI-assisted forensic tools can process large volumes of data more efficiently than human investigators. AI-generated transcripts may improve access to justice by reducing costs and increasing efficiency. Advanced image enhancement technologies may help clarify critical visual evidence. Predictive systems can assist legal professionals in organizing and analysing complex information. Consequently, the question is not whether artificial intelligence should be excluded from legal processes altogether. Rather, the challenge lies in determining the circumstances under which courts can safely rely upon AI-generated material.
Indian law currently recognizes the growing importance of electronic evidence. Judicial decisions and legislative provisions have gradually adapted to technological developments, particularly through provisions governing electronic records under the Indian Evidence Act. However, these frameworks were developed during an era in which digital evidence was primarily viewed as a record of human activity rather than a product of artificial intelligence. The emergence of AI-generated evidence therefore exposes a regulatory gap that existing legal principles may not fully address.
A central concern relates to authenticity. Traditional electronic evidence generally reflects events that actually occurred. AI-generated evidence may not. A synthetic image, voice recording, or document can appear entirely genuine despite having no connection to reality. If courts cannot reliably distinguish between authentic and artificially generated material, the evidentiary process itself may become vulnerable to manipulation.
Equally important is the issue of accountability. When human witnesses provide false testimony, they may be cross-examined. When experts present unreliable opinions, their methodology can be challenged. When documents are forged, responsibility can often be traced to identifiable individuals. Artificial intelligence complicates this framework because determining responsibility becomes significantly more difficult when evidence is generated through opaque technological systems.
The debate surrounding AI-generated evidence therefore extends beyond technology. It touches upon some of the most fundamental principles of justice, including due process, fairness, transparency, and the right to a fair trial. Courts are not merely concerned with whether evidence appears convincing; they are concerned with whether it can be verified, challenged, and trusted.
The real question before the legal system is not whether artificial intelligence can generate evidence. It undoubtedly can. The more important question is whether evidence produced, modified, interpreted, or enhanced by artificial intelligence can satisfy the standards of authenticity and reliability upon which the justice system depends. As artificial intelligence continues to evolve, the answer to that question may shape the future of evidence law in India.
USE OF LEGAL JARGON
The question of whether AI-generated evidence should be admissible in Indian courts cannot be answered solely through technological analysis. At its core, the issue is a legal one because it concerns the integrity of the judicial process, the protection of due process rights, and the evidentiary standards upon which courts rely to ascertain truth. Several legal concepts become particularly relevant when evaluating the admissibility and reliability of AI-generated material.
One of the most fundamental concepts is admissibility of evidence. Under Indian law, evidence is not admitted merely because it exists. Courts must first determine whether the material satisfies the legal requirements governing relevance, authenticity, and reliability. AI-generated evidence introduces unique challenges because it may not represent an actual event or occurrence. Unlike conventional evidence, which generally records human actions or observations, AI systems possess the ability to generate entirely new content that may appear genuine despite lacking any connection to reality.
Closely related to admissibility is the principle of relevancy. The Indian Evidence Act permits the admission of facts that are logically connected to matters in issue. However, relevance alone does not guarantee reliability. An AI-generated image, transcript, or video may appear relevant to a dispute, but the court must still determine whether the material accurately reflects the facts it purports to establish. The distinction between relevance and reliability becomes increasingly important in an era where artificial intelligence can generate convincing but inaccurate outputs.
The concept of authenticity lies at the heart of the debate surrounding AI-generated evidence. Authenticity requires proof that evidence is what it claims to be. Traditionally, authenticity may be established through witness testimony, expert verification, metadata analysis, or documentary records. AI-generated content complicates this process because synthetic material can closely imitate authentic evidence. Deepfakes, voice clones, and AI-generated documents challenge conventional methods of verification and force courts to reconsider how authenticity should be established in the digital age.
Another significant legal concern involves electronic evidence under Section 65B of the Indian Evidence Act. Judicial precedents have emphasized that electronic records must satisfy specific procedural requirements before being admitted into evidence. While Section 65B addresses the admissibility of electronic records, it was developed at a time when digital evidence primarily consisted of emails, CCTV footage, call records, and computer-generated documents. The provision does not specifically address evidence created through generative artificial intelligence, thereby creating uncertainty regarding the evidentiary treatment of AI-generated material.
The discussion also engages the principle of chain of custody, which refers to the documented history of how evidence has been collected, handled, stored, and presented before a court. Chain of custody helps ensure that evidence has not been altered or tampered with. AI-generated evidence introduces a novel challenge because the issue may not be whether the evidence was altered, but whether it ever represented reality in the first place. Establishing a reliable chain of custody becomes significantly more complicated when artificial intelligence participates in the creation, modification, or enhancement of evidentiary material.
A related concept is expert evidence. Courts frequently rely upon expert testimony when evaluating scientific, technical, or specialized issues. In disputes involving AI-generated evidence, experts may be required to explain the functioning of algorithms, assess reliability, identify manipulation, or evaluate the accuracy of AI outputs. However, this raises a further concern: if even experts struggle to fully explain how certain AI systems arrive at their conclusions, can courts confidently rely upon such evidence?
This concern leads directly to the issue of algorithmic opacity, often referred to as the “black box problem.” Many advanced AI systems generate outputs through complex computational processes that are not readily understandable to ordinary users or even their developers. The absence of transparency may conflict with fundamental principles of procedural fairness because parties must have a meaningful opportunity to challenge the evidence presented against them.
The principle of natural justice is therefore particularly significant. One of the most fundamental rules of natural justice is that no person should be condemned without an opportunity to be heard and to challenge the evidence relied upon by the opposing party. If AI-generated evidence cannot be effectively scrutinized, questioned, or explained, its use may undermine the fairness of judicial proceedings.
The debate also intersects with the constitutional guarantee of a fair trial, which forms part of Article 21 of the Constitution of India. The Supreme Court has repeatedly recognized that fairness is an essential component of due process. The admission of unreliable, unverifiable, or opaque AI-generated evidence may compromise this principle by creating uncertainty regarding the accuracy of judicial fact-finding.
Ultimately, the legal challenge posed by AI-generated evidence extends beyond questions of technological innovation. It concerns the ability of the justice system to preserve trust, transparency, and accountability in an era where machines are increasingly capable of producing information that resembles truth. As courts continue to confront emerging technologies, traditional evidentiary principles will remain indispensable in determining whether artificial intelligence can become a reliable participant in the administration of justice.
THE PROOF
The debate surrounding AI-generated evidence is no longer a theoretical discussion confined to academic circles. Courts, law enforcement agencies, forensic laboratories, and legal professionals across the world are increasingly encountering situations where artificial intelligence has played a role in creating, modifying, analysing, or interpreting evidence. The rapid expansion of these technologies has demonstrated both their potential value and their potential dangers.
One of the most significant challenges arises from the development of deepfake technology. Modern artificial intelligence systems are capable of generating highly realistic images, videos, and audio recordings that can be extremely difficult to distinguish from genuine content. A fabricated video may depict an individual making statements that were never spoken. A synthetic audio recording may replicate a person’s voice with remarkable accuracy. In legal proceedings, where credibility and authenticity are essential, such technologies present a serious threat to the evidentiary process.
The danger posed by deepfakes is not merely that false evidence can be created. The greater concern is that genuine evidence may also become vulnerable to challenge. If courts become aware that realistic digital fabrications can be produced with relative ease, parties may begin questioning the authenticity of legitimate electronic evidence. This phenomenon, sometimes referred to as the “liar’s dividend,” allows individuals to dismiss genuine recordings by claiming they were artificially generated. Consequently, artificial intelligence has the potential to undermine confidence in both false and authentic evidence.
Another concern involves AI hallucinations. Unlike traditional software that follows predetermined instructions, generative AI systems produce outputs based on patterns identified within training data. As a result, they occasionally generate information that appears plausible but is factually incorrect. Numerous incidents have demonstrated AI systems producing fictional citations, inaccurate legal authorities, fabricated facts, and non-existent sources. If similar inaccuracies appear within evidence relied upon during judicial proceedings, the consequences could be severe. A legal system built upon factual accuracy cannot afford to rely unquestioningly upon outputs that may contain unverifiable information.
Artificial intelligence is also increasingly used to generate transcripts, summarize documents, analyse digital records, and process large volumes of information. These capabilities can significantly improve efficiency. However, efficiency does not necessarily guarantee accuracy. Errors in transcription, contextual misunderstandings, and incorrect interpretations may influence the presentation of evidence in ways that are not immediately apparent. The risk is particularly significant where AI-generated outputs are accepted without adequate human verification.
The growing use of AI-assisted forensic tools presents another challenge. Artificial intelligence is now capable of analysing facial images, identifying behavioural patterns, processing surveillance footage, and assisting in criminal investigations. While these tools may enhance investigative capabilities, concerns remain regarding algorithmic bias, transparency, and reliability. Research has repeatedly demonstrated that certain AI systems may produce different levels of accuracy across demographic groups. If such systems influence legal outcomes, questions concerning fairness and equality before the law inevitably arise.
International developments illustrate that courts are beginning to approach AI-generated evidence with caution. Legal systems increasingly recognize that technological sophistication alone does not guarantee reliability. Judges and policymakers are paying greater attention to issues such as explainability, transparency, verification, and procedural fairness. The emerging consensus appears to be that artificial intelligence may assist the judicial process, but its outputs cannot be accepted without scrutiny.
India faces a particularly significant challenge because existing evidentiary frameworks were developed before the emergence of generative artificial intelligence. While provisions governing electronic records provide valuable safeguards, they do not directly address situations involving synthetic evidence, AI-generated content, or algorithmic decision-making. This creates uncertainty regarding how courts should evaluate evidence that has been materially influenced by artificial intelligence.
At the same time, it would be inaccurate to portray AI-generated evidence as entirely problematic. Artificial intelligence possesses the potential to improve access to justice, enhance efficiency, reduce costs, and assist courts in managing increasingly complex disputes. AI-powered tools can help identify relevant information, organize large datasets, and support legal analysis. The challenge therefore is not whether artificial intelligence should be used within legal systems, but how such use can be regulated without compromising fairness and reliability.
The evidence available today points toward a clear conclusion. Artificial intelligence can be an extraordinarily powerful tool within the justice system, but it is not yet a substitute for human judgment. Courts are institutions founded upon trust. Before AI-generated evidence can be relied upon in judicial proceedings, the legal system must ensure that such evidence satisfies the same standards of authenticity, transparency, and reliability that have governed admissible evidence for generations. The future of AI in courtrooms will therefore depend not upon technological capability alone, but upon the ability of law to preserve confidence in the pursuit of truth.
ABSTRACT
Artificial intelligence is rapidly transforming the way information is created, analysed, and presented. As AI systems become increasingly sophisticated, they are beginning to influence legal processes through the generation, enhancement, interpretation, and organization of evidence. From AI-generated transcripts and facial recognition technologies to synthetic media and deepfakes, artificial intelligence is no longer merely a technological innovation—it is becoming a potential participant in the administration of justice.
This development raises a fundamental legal question: can courts rely upon evidence generated or significantly influenced by artificial intelligence? While AI offers considerable benefits in terms of efficiency, accuracy, and data processing, it also introduces concerns relating to authenticity, reliability, transparency, and procedural fairness. The ability of AI systems to create highly convincing but potentially inaccurate content challenges traditional evidentiary principles that have long governed judicial proceedings.
This article examines the admissibility of AI-generated evidence in India through the lens of evidence law, constitutional safeguards, and emerging technological realities. It analyses the legal challenges associated with AI-generated material, the limitations of existing evidentiary frameworks, and the need for regulatory safeguards capable of preserving trust within the justice system. The article argues that while artificial intelligence may serve as a valuable tool in judicial processes, its outputs must remain subject to rigorous scrutiny before they can be relied upon as evidence in courts of law.
IMPORTANT CASE LAWS
1. Anvar P.V. v. P.K. Basheer (2014)
This landmark judgment transformed the treatment of electronic evidence in India. The Supreme Court held that electronic records must comply with the requirements of Section 65B of the Indian Evidence Act in order to be admissible. The case remains highly relevant because AI-generated evidence would likely fall within the broader category of electronic evidence and therefore be subject to similar evidentiary safeguards.
2. Arjun Panditrao Khotkar v. Kailash KushanraoGorantyal (2020)
The Supreme Court reaffirmed the mandatory nature of Section 65B certification for electronic evidence. The judgment emphasized the importance of authenticity and procedural compliance, principles that become even more significant when evaluating AI-generated material.
3. State of Maharashtra v. Dr. Praful B. Desai (2003)
In this case, the Supreme Court recognized the validity of technological advancements in judicial proceedings and permitted evidence to be recorded through video conferencing. The judgment demonstrated the judiciary’s willingness to adapt to technological change while maintaining procedural safeguards.
4. Tomaso Bruno v. State of Uttar Pradesh (2015)
The Court highlighted the growing importance of electronic evidence in modern criminal investigations. The judgment reinforced the principle that technological evidence can play a crucial role in the administration of justice when properly authenticated and verified.
5. Justice K.S. Puttaswamy v. Union of India (2017)
While primarily concerned with privacy rights, this landmark judgment emphasized informational privacy and individual autonomy. The principles established by the Court remain relevant because AI-generated evidence often relies upon extensive data collection and algorithmic processing that may affect privacy rights.
CONCLUSION
The integration of artificial intelligence into legal systems represents one of the most significant transformations in the history of evidence law. For centuries, courts have relied upon witnesses, documents, expert testimony, and physical evidence to establish facts and resolve disputes. Today, artificial intelligence challenges traditional assumptions by introducing a new category of evidence that may be generated, modified, interpreted, or enhanced by machines.
The central concern is not whether artificial intelligence can assist the administration of justice. In many respects, it already does. AI systems can process vast quantities of information, improve efficiency, support investigations, and assist legal professionals in managing complex cases. The real challenge lies in determining whether courts can safely rely upon evidence produced by technologies whose reasoning may not always be transparent, explainable, or verifiable.
In my view, the admissibility of AI-generated evidence should not depend solely upon technological sophistication. It should depend upon whether the evidence satisfies the fundamental legal principles that have long governed judicial proceedings: authenticity, reliability, transparency, and fairness. These principles exist not to resist innovation but to protect the integrity of the justice system itself.
The future of evidence law in India will likely involve increasing interaction between human judgment and artificial intelligence. However, trust cannot be automated. Before AI-generated evidence is permitted to influence legal outcomes, courts must ensure that such evidence can withstand the same level of scrutiny applied to traditional forms of proof. The pursuit of justice demands nothing less.
Ultimately, the question is not whether artificial intelligence will enter Indian courtrooms. It already has. The more important question is whether the legal system can develop safeguards that allow innovation to coexist with due process, ensuring that technological advancement strengthens justice rather than undermines it.
FAQS
Q1. What is AI-generated evidence?
AI-generated evidence refers to information, images, videos, audio recordings, transcripts, analyses, or other materials that are created, modified, enhanced, or interpreted using artificial intelligence systems.
Q2. Is AI-generated evidence admissible in Indian courts?
Indian law does not currently contain specific provisions dealing exclusively with AI-generated evidence. Its admissibility would generally depend upon existing rules governing electronic evidence, authenticity, reliability, and compliance with evidentiary requirements.
Q3. Why is AI-generated evidence controversial?
AI-generated evidence raises concerns regarding deepfakes, fabricated content, algorithmic bias, lack of transparency, and difficulties in verifying authenticity and accuracy.
Q4. What is the biggest challenge associated with AI-generated evidence?
The primary challenge is ensuring that the evidence is reliable, authentic, and capable of being independently verified before it influences judicial decisions.
Q5. Can artificial intelligence replace judges in evaluating evidence?
No. Artificial intelligence may assist legal processes, but judicial decision-making ultimately requires human judgment, legal reasoning, and procedural fairness that cannot currently be replaced by automated systems.
