Artificial Intelligence in the Identification of the Deceased: Forensic Innovations, Evidentiary Reliability, and Legal Implications in India

INTRODUCTION

Identification of the deceased takes a centre stage in the area of forensic science, and it is evidentiary and humanitarian in the criminal justice system at the same time. Determining cause and manner of death, criminal liability, missing persons investigation, and restoration of dignity to the deceased, is often a requirement of establishing identity after death which provides closure to the families of the deceased. Historically, this identification has been based on the conventional modalities of forensic identification, such as visual recognition, fingerprints, dental records, DNA profiling of the body and forensic anthropology.¹ These traditional methods have increasingly over the last few years been augmented (and in other cases replaced) by tools that employ AI and claim improvements in accuracy, speed and scalability of forensic identification.²

The current application of AI in the identification of the deceased is now a wide range of forensic processes that include facial recognition of post-mortem images, craniofacial reconstruction, algorithm-aided DNA interpretation, sex and age profiling using skeletal evidence, and the evaluation of radiological and pathological evidence.³ The technologies are especially useful when it comes to the situation of a mass disaster, advanced decomposition, or fragmented remains, in which the conventional techniques are usually subject to significant limitations.⁴ The academic literature of the global community is recorded as more and more confident in machine-learning and deep-learning models to profile biologically and attribute identities and often claim them as objective, scientifically better than the human mind.⁵

In India, the implementation of AI in forensic and policing organizations is experienced against the background of wider state-led digitalisation efforts in the criminal justice system. The presence of a policy orientation to use AI to make police, forensic labs, and courts more efficient, accurate, and interoperable is evident in government reports and institutional publications.⁶ At the same time, new legislative changes have been implemented recently, and their first and most significant event is the passage of the BSA which promotes the need to focus on expert evidence and scientific opinion in criminal cases.⁷ The meeting of the technological innovation and legal reform makes the issue of AI-aided forensic identification especially pertinent to the modern Indian criminal procedure.

Nevertheless, the legal consequences of using AI in identification of a dead body are not thoroughly theorised in the Indian academic and jurisprudential field. However, as much as the question of admissibility and reliability of forensic evidence has been long-standing in courts,⁸ the emergence of algorithmic decision-making has qualitatively new considerations. AI systems tend to be “black boxes”, producing outputs that are hard to interpret, criticise or substantively investigate using conventional approaches to cross-examination.⁹ This obscurity makes it difficult to determine traditional principles of evidentiary practice of expert testimony, probative effect, and judicial evaluation of scientific reliability.¹⁰ Moreover, any mistakes or biases incorporated into AI models can have far-off implications, such as incorrect identity, criminal investigation diversion, and loss of procedural fairness.¹¹

These issues become constitutional in the Indian context. There is no technical exercise in the identification of the deceased as it touches upon the right to a fair trial, the presumption of innocence and the right to dignity provided in Article 21 of the Constitution of India.¹² There is a danger of eroding these protections in case an uncritical dependence on AI-produced identification evidence is used in court and the judges cannot assess the scientific quality and restriction of such tools independently. Jurisdiction comparative jurisprudence of jurisdictions like the United States and the United Kingdom also reflects judicial discomfort in cloudy scientific methods or scientific methods that lack a sufficiently validated basis, especially when they have a direct bearing on criminal culpability.¹³

It is on this background that the current paper aims to explore the application of AI in the identification of the deceased using a forensic and socio-legal perspective. It contends that although AI-guided technologies have irrefutable advantages to forensic practice, they must be carefully introduced into the Indian criminal justice system and thoroughly handled in legal terms. In the absence of explicit standards to regulate admissibility, validation, and judicial scrutiny, AI-assisted identification might undermine, and not improve the quest of forensic justice.

Objectives of the Study

1.  To examine the emerging applications of AI in the forensic identification of deceased persons.

2.  To analyse the admissibility and evidentiary value of AI-assisted forensic identification under Indian law.

3.  To evaluate the reliability, transparency, and explainability of AI-driven identification tools.

4.  to assess the extent to which existing legal doctrines enable effective judicial scrutiny and cross-examination of algorithmically generated forensic outputs.

5.  To situate the use of AI in deceased identification within constitutional safeguards.

6.  To undertake a comparative analysis of foreign jurisprudence and regulatory approaches relating to the use of algorithmic and scientific evidence in criminal proceedings.

7.  To propose legal and institutional recommendations for regulating the use of AI in forensic identification in India.

RESEARCH METHODOLOGY

PublMed, Web of Science, Scopus, and Google Scholar were searched through a systematic search, and it was complemented by systematic manual reviews of major forensic journals and grey literature. The review included the studies that examined the use of AI in forensic odontology and other forensic activities, published in the previous decade. There were predefined inclusion and exclusion criteria used and the duplicate records were eliminated. Relevance was checked through full-text reviews.

The current research project is based on a qualitative approach to the doctrinal and socio-legal study, which is based on the examination of primary and secondary sources.

a) Primary sources:

Within the context of the admissibility, dependability, and assessment of expert forensic and scientific evidence, the study examines primary legal material such as the BSA applicable provisions of criminal procedure and judicial rulings of the Supreme Court of India and High Courts. The treatment of algorithmic and scientific evidence has also been analyzed based on selected foreign judicial decisions that have occurred in the jurisdictions of the United States and the United Kingdom to offer a comparative understanding of how they are dealt with.

b) Secondary sources:

The secondary sources include academic research articles, interdisciplinary literature in forensic science, institutional and governmental reports as well as policy documents which question the use of AI to identify the deceased, its use in facial recognition, DNA interpretation, forensic anthropology, and post-mortem examination. These resources have been used to determine the technological changes, the reliability of evidence, and the socio-legal aspects of AI-assisted forensic identification in the Indian criminal justice system.

Scope and Limitation

The research paper is limited to exploring the use of Artificial Intelligence in detecting dead individuals in the fields of forensic science and criminal justice in India. In content, it concentrates on AI-assisted technologies like post-mortem facial recognition, DNA testing, forensic anthropology, and post-mortem photography, and how they are admissible in the criminal trial. The Indian evidentiary and constitutional framework, specifically, the BSA treatment of expert evidence by the court, and constitutional protection under Article 21 of the Constitution of India are subjected to legal analysis. Only to the limited degree that is important in placing Indian law in context, selectively drawing upon jurisprudence in the United States and the United Kingdom, is a comparative perspective employed.

The research is literature-based and doctrinal and not centering on empirical validation of AI technologies. It uses published judicial rulings, the scholarly literature, and internal resources, and this information might not be a complete representation of confidential or emerging forensic practices. The conclusions have thus been conditional on the current legal frameworks and reported practices, and may be subject to future doctrinal and technological changes.

Literature Review

Title	Type	Journal / Publisher	Purpose	Key Findings	Research Gap
Artificial Intelligence in the Identification of the Deceased: A Narrative Review with Implications in Forensic Science14	Journal Article	Behavioral Sciences & the Law (Wiley)	To review AI-based techniques used in identifying deceased individuals across forensic domains	Highlights growing use of machine learning in facial reconstruction, skeletal analysis, and biological profiling with claims of improved accuracy	Does not examine legal admissibility, judicial scrutiny, or constitutional implications of AI-based identification evidence
A Narrative Review on Application of Artificial Intelligence in Forensic Sciences15	Journal Article	Journal of Indian Oral Health (2025)	To analyse applications of AI in various forensic sub-fields including pathology and identification	Notes efficiency gains and reduced human error in forensic analysis using AI	Lacks discussion on evidentiary reliability standards and legal safeguards in criminal trials
Revolutionizing Forensic Science: The Role of Artificial Intelligence in Evidence Analysis16	Journal Article	International Journal of Interdisciplinary Approaches in Psychology	To examine AI tools used in evidence processing and forensic decision-making	Emphasises AI’s role in speeding up forensic analysis and enhancing pattern recognition	Treats AI outputs as neutral and objective, without addressing risks of bias or judicial evaluation
The Future of Artificial Intelligence in Forensics: Advancements, Challenges, and Ethical Considerations17	Research Paper	Independent Academic Publication	To explore future potential, challenges, and ethical issues of AI in forensic science	Identifies ethical risks such as bias, privacy concerns, and over-reliance on algorithms	Does not connect ethical concerns with evidentiary admissibility or constitutional protections
Artificial Intelligence in Forensic Science18	Book Chapter	IGI Global	To provide an overview of AI tools and techniques in forensic science	Describes AI applications in identification, crime reconstruction, and forensic databases	Focuses on technological aspects without engaging with legal standards or judicial oversight
Digital Detectives: Exploring the Integration of Artificial Intelligence in Indian Forensic Medicine19	Research Paper	IP International Journal of Forensic Medicine and Toxicological Sciences	To examine integration of AI in Indian forensic medicine and pathology	Highlights institutional adoption of AI in forensic medicine in India	Does not analyse legal preparedness or admissibility of AI-generated forensic opinions
Use of Artificial Intelligence in Aiding Forensic & Legal Enforcement: Opportunities and Ethical Implications20	Journal Article	Indian Police Journal (BPR&D)	To study AI use in policing and forensic assistance in India	Advocates AI adoption for identification, investigation, and efficiency	Lacks critical analysis of evidentiary reliability and safeguards in criminal adjudication
Digital Courts: Vision & Roadmap (Phase III)21	Government Vision Document	e-Committee, Supreme Court of India	To outline India’s vision for technology-enabled courts and justice delivery	Emphasises algorithmic transparency, data governance, and trust in digital justice	Silent on forensic-specific AI evidence and standards for judicial scrutiny
Press Information Bureau Release on AI in Judiciary / Criminal Justice22	Government Press Release	Government of India (PIB)	To communicate governmental vision on AI-driven justice reforms	Signals institutional push towards AI adoption across justice institutions	Does not address forensic AI admissibility or risks in criminal trials
INTERPOL Disaster Victim Identification Guide23	Institutional Report	INTERPOL	To standardise identification procedures in mass disasters	Emphasises accuracy, chain of custody, and reliability in victim identification	Does not account for AI-driven identification tools or their legal implications
Postmortem Biometric Identification Using Artificial Intelligence24	Research Article	IEEE / ArXiv	To study AI-based biometric identification of deceased persons	Demonstrates technical feasibility of post-mortem biometric recognition	Ignores legal admissibility, cross-examination, and procedural fairness concerns

OVERALL RESEARCH GAP

Available sources on AI-aided forensic identification of the deceased generally pay attention to technological efficiency and ethical issues, but little attention is paid to evidentiary admissibility, judicial review, and constitutional protection in the Indian context. Though the Indian policy frameworks are supportive of the integration of AI in criminal justice institutions, the standards of assessing AI-generated identification evidence under the BSA and safeguarding interests of fair trial and dignity under Article 21 are not prescribed. Even though comparative jurisprudence has established reliability and gate keeping principles of scientific evidence, they were not systematically implemented to AI-assisted deceased identification in India, demonstrating a significant lack of links between forensic innovation and legal regulation.

SOCIO-LEGAL ANALYSIS

FORENSIC IDENTIFICATION OF THE DECEASED: FROM HUMAN EXPERTISE TO ALGORITHMIC DECISION-MAKING

Traditionally, the process of identifying the deceased has been based on human knowledge, wherein the interpretation of human physical remains and other contextual evidence has depended on the professional judgment of forensic pathologists, anthropologists, odontologists and investigators. Conventional approaches, like visual identification, comparison of fingerprints, dental records, and DNA profiling, were moderated by expert judgment, were susceptible to a set of protocols of confirmation and court examination.²⁵ This anthropocentric paradigm placed responsibility and accountability directly in the hands of the forensic expert whose opinion may be subject to cross-examination and admissible in court in terms of experience, practice, and method.²⁶

The presentation of AI in the field of forensic identification is a major shift of this paradigm. AI-powered systems have become helpful, and in other cases significantly contribute to the identification of individuals who are deceased, using automated facial recognition of post-mortem photos, the algorithmic analysis of DNA profiles, skeletal analysis, and to the radiological pattern recognition.²⁷ The technologies work on the basis of establishing statistical correlations in large volumes of data, and they generate probabilistic results that can be faster than human speed and, under controlled conditions, more accurate.²⁸ With the growing presence of such systems in forensic literature as objective and helpful in terms of efficiency, the decision-making process is slowly being overtaken by algorithmic inference, where the expert in question no longer holds the ultimate authority.²⁹

This has significant socio-legal ramifications. In contrast to expert opinion as used historically, AI systems can usually be opaque black boxes, that is, the rationale that supported a conclusion cannot be intuitively accessed or explained in all aspects, even to the forensics specialist using the tool. The work of the forensic expert runs a risk of turning into a mediator between the smooth operation of the algorithms rather than an autonomous assessor who is able to report the results of the algorithmic process without having full understanding of its internal mechanisms and the possible causes of the abuse. Such a change casts underlying doubts of the question of accountability, epistemic authority and the seat of expertise in the adjudication of crimes- questions that the prevailing evidentiary dogmas were not meant to answer.³⁰ This shift in the human judgment towards algorithmic decision-making thus preconditions further issues of admissibility, judicial review, and procedural fairness which are discussed in the following sections.³¹

B. Admissibility of AI-Assisted Forensic Evidence under Indian Evidence Law

The issue of the admissibility of AI-aided forensic identification evidence in India has to be discussed in the context of the rules that define the expert opinion and scientific evidence which is now centralized under the BSA. It is a fundamental rule retained by the BSA that courts are permitted to resort to expert opinion in those cases where the relevant subject matter is one of specialised knowledge beyond ordinary judicial competence, and at the same time accords the courts power to give discretion over the reliability and probative value of such expert opinion. With the forensic identification of the deceased, AI-based outputs usually enter into the system with the assistance of human experts, including forensic scientists or pathologists, which use algorithmic tools to support their decisions. This brings up the prior question, whether AI outputs are expert opinion or just an input to expert opinion, which has important evidentiary implications.

Indian courts have always believed that expert evidence is not binding, but only advisory and that it has an evidentiary value based on how well the methodology to be used and the logic provided are sound. State of H.P. v. Jai Lal,³² the Supreme Court pointed out that the courts should be convinced that an opinion of an expert should be formed by recognised principles and tested methods, and not due to unproven assumptions. On the same note, in Ramesh Chandra Agrawal v. Regency Hospital Ltd. the Court emphasized that the expert testimony should pass the test of judicial examination regarding the credibility, methodology, and reliability with the existing scientific standards as provided in a particular case.³³ Through these principles being applied to AI-assisted forensic identification, there is revealed a doctrinal collision where the courts can question the reasoning behind the expert in traditional forensic practice but AI systems frequently produce probabilistic outcomes without an easily explainable logic.

That obscurity makes it difficult to apply the available evidentiary protections. With the BSA, the courts should also look at the qualification of the expert in addition to the foundation of the opinion tendered. In cases where such expert heavily depends on an AI model, the training data, error rates, or logic of decision-making of which are not fully disclosed to the expert, the court is severely limited in its ability to independently determine reliability. The danger is that outputs of algorithms can be given unwarranted evidentiary weight because they are scientifically looking, as opposed to being provable. This deference would be incompatible with established status quo that scientific evidence should be critically assessed and it should not be accepted just because it is framed in the terms of technical or technological sophistication.³⁴

Based on this, the BSA does not specifically comment on the issue of AI-generated forensic evidence, but the fundamental principles of it require a careful and stringent approach. The identification of the deceased using AI can only be aligned with Indian evidence law when the courts consider algorithmic tools as supporting technologies that can be rigidly validated, but not as the sources of expert opinion themselves. The lack of clear statutory guidance concerning this matter can also be seen as a strong indication that judicial-developed standards must be applied to the AI-monitored forensic identification evidence to make sure that it meets the conventional relevance, reliability, and fairness criteria.

C. Judicial Scrutiny, Cross-Examination, and the Problem of Algorithmic Opacity

A crucial protection that regulates the admissibility of an expert testimony in a criminal case lies in the ability of the court to place the expert testifying under the close examination of justice by the court mainly by cross-examination and adversarial testing. The Indian law of evidence has long held the opinion of experts as a relatively weak evidence, and thus should be appraised rather than accepted without a second thought.³⁵ Adversarial process helps courts to examine what is the qualification of the expert, what methodology is being used and how data is related to conclusions.³⁶ Nonetheless, the growing use of AI in forensic identification creates a structural problem to the process, where algorithmic systems tend to work in a way that is not easily interrogated.

Such forensic tools which are driven by AI, especially those which apply machine learning and deep-learning models, are often “black boxes”, with the internal decision-making logic being obscure even to their creators. The outputs of such systems when applied to recognize deceased people (whether by facial analysis, skeletal inference, or DNA analysis) are usually probabilistic, based on complicated statistical associations.³⁷ Practically, the forensic specialist that provides such evidence might not be capable of demonstrating the manner in which a specific result was obtained besides stating that it was created using an authenticated algorithm. This goes a long way to undermine the effectiveness of the cross-examination process because the other party is not allowed the chance to inquire on the possible areas of error, prejudice, or other possible interpretations.

This particular type of evidentiary opacity has been forced to be confronted by Indian courts. In State of H.P. v. Jai Lal, the Supreme Court repeated its statement that expert evidence must be subject to judicial testing and thus, the need to avoid blind application of expert knowledge.³⁸ Likewise, the Court has reiterated on several occasions that the use of scientific evidence cannot be accorded conclusiveness unless its methodology has been proven to be reliable and transparent. Applying AI-generated evidence of forensic identification, which cannot be explained satisfactorily, has the potential to go against this principle and change the role of the court to the machine.³⁹

The socio-legal implications of the algorithmic opaqueness, therefore, form a twofold menace: undermining the procedural rights of the accused due to the restriction of a successful cross-examination, as well as the judicial independence due to the establishment of obedience towards the technological power.

In the area of the identification of the deceased, the socio-legal implications of algorithmic obscurity are manifold, since they not only undermine the procedural rights of the accused by means of effectively cross-examining, but also impair the judicial independence by creating a sense of deference to technological power. Such obscurity is, therefore, a significant challenge to the fidelity of criminal adjudication. The issues highlight the urgency of revisiting the approaches to the process of evaluating AI-enabled forensic evidence in courts, which cannot be delegated to technological verification only.

D. Constitutional Safeguards: Fair Trial, Human Dignity, and Due Process

The use of AI in the forensic identification of the deceased involves the core constitutional protection in the Article 21 to the Constitution of India that ensures the rights to life, personal liberty, and a fair trial under the due process of justice that is both just, fair, and reasonable.⁴⁰ The Supreme Court has always construed Article 21 to include the rights of the accused and other wider interests of procedural fairness, human dignity, and substantive due process in criminal trials.⁴¹ When applied to the concept of AI-aided forensic identification, these constitutional protections take on an even more critical importance here, since the results of the algorithms can have a direct impact on the course of investigation, the establishment of identity, and ultimately criminal responsibility.

Indian constitutional jurisprudence has noted that the application of scientific methods in criminal cases should not override on an individual freedom, dignity as well as fairness. In Selvi v. State of Karnataka,⁴² the Supreme Court warned of the uncontrolled application of the science in impairing the privacy of the mind and the independence of the evidentiary processes, and the constitutional safeguards could not be diluted by technological progress. In spite of the fact that the topic of testimonial compulsion was discussed in the case of Selvi, its arguments are still informative in the current context: AI-based forensic identification, when considered as infallible and beyond questioning, poses a risk to the right of the accused to challenge the evidence and the responsibility of the court to provide a fair trial. This balance is further complicated by the opacities and probabilistic characteristics of algorithmic identification because the parties impacted by it are not able to make meaningful sense of the evidence on which they rely or challenge it.

In addition to the right of the accused, the identification of the deceased engages the right to dignity and the Supreme Court has determined that this right will exist even after death and apply to the treatment of human remains and the interests of family members left alive. Incorrect or insufficiently examined AI-aided identification can lead to misidentification, deprivation of last rites, or extended doubt on the part of the family, which violates the interests of dignity guaranteed by Article 21. The constitutional duty of the State to promote fairness and accuracy in criminal proceedings, therefore, applies to protecting the dignity of the deceased and the social trust to the forensic institutions.⁴³

Based on this, the constitutional structure from the foregoing perspective requires that AI-based forensic identification must be placed under a more procedural protection, as opposed to a less vigilant review. The courts should make sure that this kind of evidence should be transparent, challengeable, and in line with the interests involved and that is in line with the insistence of the Supreme Court that technology tools cannot replace constitutional values.⁴⁴ The lack of clear statutory or judicial interpretation of these protects highlights the importance of dynamically interpreting the current constitutional principles so that innovation in forensic science does not undermine the original protection of fairness, dignity, and due process in the criminal justice system.

E. Comparative Jurisprudence on Scientific and Algorithmic Evidence

The use of comparative legal approaches to scientific and algorithmic evidence can be used to show that there is a uniform judicial reticence to provide automatic credibility to technologically advanced evidence, which reinforces the role of judicial gatekeeping and reliability testing. In the U.S, the ruling of the Supreme Court in Daubert v. Merrell Dow Pharmaceuticals, Inc. defined a list of criteria, which were not exhaustive, as testability, peer review, known or possible rate of errors, and the overall acceptability by the courts as a guideline in determining the admissibility of scientific evidence.⁴⁵ Later cases have supported the position of the trial judge as a gatekeeper, whose duties are to reject evidence that cannot be proved to be among those that are reliable, regardless of their technical complexity. The methodological transparency and error evaluation components of Daubert, though not specifically focused on AI, have been called upon more and more in attempts to put algorithmic and forensic technologies on trial, especially when probabilistic models are involved in fact-finding.⁴⁶

Courts of the United Kingdom have also been quite reserved in accepting expert and scientific evidence. In R v. Dlugosz the Court of Appeal cautioned that expert testimony dressed up as scientific authority should not be received without question in a very uncritical manner without the necessary knowledge of the limitations and uncertainties of such evidence being properly informed to the jury and the judge.⁴⁷ In more recent times the use of opaque, or inadequately verified forensic methods has been within the spotlight of British courts and law reform bodies who have emphasized the need to have a clear description of methodology and margins of error.⁴⁸ The philosophy of this method is seen in a wider judicial concern that scientific innovation still be subservient to notions of fairness and rational judging.

These comparative models are educative to the Indian context especially with regards to AI-assisted recognition of the deceased. In both the U.S. and the U.K. jurisprudence, the idea that technological sophistication can take the place of legal scrutiny is disapproved of and that explainability, validation and testing adversarity are the conditions of admissibility.⁴⁹ By comparisons, Indian evidence law does not give clear guidelines on the evaluation of algorithmic or probabilistic forensic evidence, and courts must make recourse to the general principles of expert opinion. The lack of an organized gatekeeping system increases the likelihood that AI-generated identification evidence can be accepted without further questioning of its scientific and limitations.

Comparative jurisprudence, therefore, does not propose the mass importation of foreign teachings into Indian law, but it also has provision to the useful normative advice. This demand on judicial regulation of scientific evidence, scientific method transparency and recognition of uncertainty can serve as a model to work on Indian standards that balance forensic innovation and constitutional and evidentiary protection.⁵⁰ These principles in the context of the AI-assisted identification of the deceased are essential to state that the Indian judicial practice has to be developed to guarantee that algorithmic tools contribute to the codification of the integrity of the criminal adjudication, rather than to its decline.

SOCIO-LEGAL SYNTHESIS: TECHNOLOGY, TRUST, AND THE CRIMINAL JUSTICE SYSTEM

The involvement of the AI in the process of forensic identification of the deceased is not just the change in technology; it is the beginning of an extensive change in the relationship between science, law, and the values of people in response to the criminal justice system.⁵¹ The legitimacy of forensic identification in the past has been established through the perceived objectivity of empirical procedures and legal checks and balances and through the adversarial and judicial review. Identification using AI breaks this balance by adding algorithmic mechanisms that might seem scientifically legitimate but have insufficient transparency or challenge.⁵² Once the courts, investigators or forensic entities give undue deference to the results of the algorithms, a danger arises wherein technological efficiency is both confused with the idea of evidentiary reliability, hence compromising the question of judicial accountability.

Socio-legally, the validity of forensic evidence should not be based on technical accuracy but on the ability of legal systems to explain, analyze, and defend the usage of such evidence to the parties impacted such as the accused and the bereaved families. When the identification of a deceased is concerned, mistakes and lack of clarity can have consequential consequences: they may influence the investigation in the wrong direction, convict the wrong person, or deny the family its integrity and closure. The lack of regulation or transparency surrounding the use of AI therefore risks undermining trust in the forensic institutions and the courts especially when the concerned parties feel that they are being influenced by the machine and not by reasoning and the law.⁵³

Therefore, the socio-legal issue with AI-assisted forensic identification is the issue of balancing innovation with institutional confidence. The criminal justice system needs to ensure that the use of algorithmic tools should be subservient to legal justifications, highly scrutinized and in line with the constitutional ideals of fairness and dignity. Without these precautions, the increasing dependence on AI poses a threat of turning forensic science as a supplement to justice to a new and more obscure type of power. Such a synthesis highlights the importance of ensuring that AI becomes embedded in, and not overlaying the normative structure of criminal adjudication, which is an imperative that guides the issues and suggestions which are presented in the next section.

RECOMMENDATIONS

1. Judicially Enforced Reliability Gatekeeping

To require evidence of testability, validation, peer review, known error rates, and limits of applicability, Indian courts should seek to apply a structured reliability inquiry of AI-assisted forensic evidence based on comparative principles. This type of gatekeeping is harmonious with the current jurisprudence which requires the critical review of expert evidence.⁵⁴

2. Mandatory Explainability and Disclosure Obligations

Admissibility of evidence based on AI-assisted identification should depend upon the availability of the methodology of the model, the source of the training data (where practicable), the validation research and full performance statistics. In some situations in which exhaustive disclosure is unworkable, the judiciary must regard such AI-generated outputs as corroborative, but not conclusive, evidence.⁵⁵

3. Statutory Clarification under Evidence Law

Authoritative judicial clarification through legislation must refer to AI systems as auxiliary equipment, and it is the human expert who must clarify, contextualise, and justify conclusions. With the help of such demarcation, the rigor of adversarial testing and the independence of judicial processes are maintained.⁵⁶

4. Accreditation and Standard-Setting for Forensic AI

Accreditation of AI tools applied in forensic identification, periodic audits, bias evaluation, and record of change should be a mandatory national standard.⁵⁷ The accreditation authority must match the international best practices but must be responsive to the Indian situation.⁵⁸

5. Capacity Building and Judicial Training

Judges, prosecutors, defence counsel and forensic experts should be trained specifically on the fundamentals of AI, limitations and evidentiary risks to make informed questioning and avoid automation bias.⁵⁹

6. Dignity-Centred Protocols for Deceased Identification

Accuracy, transparency, and family communication should be prioritised in protocols considering posthumous dignity in Article 21.⁶⁰ The identifications aided by AI must be supported by other independent techniques prior to finalisation in delicate cases.

CONCLUSION AND WAY FORWARD

The rising adoption of the AI into the forensic identification of the deceased is a notable technological advancement that can increase the accuracy, efficiency, and responsiveness of the criminal justice system. The AI-assisted tools, when used sensibly, should assist forensic specialists in solving difficult identification problems that emerge due to mass disasters, advanced decay or fragmentation of the remains. But as this paper has indicated, there is a danger of losing the tried and tested principles of evidentiary processes, fairness of procedures, and constitutional protection as a result of the uncritical application of these technologies in the Indian criminal justice system. The lack of clear guidelines on the admissibility, validity, and judicial review of AI-generated forensic identification evidence introduces a normative vacuum where the power of technology can become bigger than the power of legal arguments.

The doctrinal basis to govern scientific and expert evidence is already established in Indian evidence law, specifically in the BSA but it needs to be applied intentionally to apply the principles to the unique issue of algorithmic decision making. Comparative jurisprudence brings out the importance of the fact that it is impossible to replace reliability, transparency and adversarial testing by scientific sophistication and the courts cannot abandon their gatekeeping functions even when dealing with complex technologies. Constitutionally, the usage of AI in the identification of the deceased would touch upon the right to the fair trial and dignity of both the accused and the deceased, requiring stronger questioning as opposed to defersiveness.

In the future, the ethically sound approach to the introduction of AI into forensics in India needs a concurrent development of the legal doctrine, the capacity of the institutions, and the forensic norms. Judicially designed reliability tests, clarification of the evidentiary status of the AI-assisted outputs in statutes, accreditation of the AI forensic instruments and long-term capacity building of the judges and forensic professionals are critical in the reinforcement of the application of technology to aid in the pursuit of justice, as opposed to undermining it. The solution to the nexus of technological advancement and criminal justice, however, does not rest in opposing changes and adapting to them, but rather in entrenching AI deep within the normative and constitutional fabric of criminal adjudication, in such a way that forensic science remains the servant of fairness, accuracy, and human dignity in the administration of justice.

*****

Footnotes

1. David L. Faigman et al., Modern Scientific Evidence: The Law and Science of Expert Testimony 1–12 (2023).

2. Vaibhavi Tiwari, The Future of Artificial Intelligence in Forensics: Advancements, Challenges, and Ethical Considerations 3–6 (2024).

3. A. Kumar et al., Artificial Intelligence in the Identification of the Deceased: A Narrative Review with Implications in Forensic Science, 27 Behav. Sci. & L. 1, 2–4 (2024).

4. INTERPOL, Disaster Victim Identification Guide 45–52 (2018).

5. A. Ross & K. Jain, Postmortem Biometric Identification, 43 IEEE Trans. Pattern Analysis & Mach. Intelligence 1, 3–5 (2021).

6. Bureau of Police Research & Development, Use of Artificial Intelligence in Aiding Forensic & Legal Enforcement 12–18 (Indian Police Journal, Jan–June 2025).

7. Bharatiya Sakshya Adhiniyam, 2023, §§ 39–42.

8. State of H.P. v. Jai Lal, (1999) 7 SCC 280, 286–87.

9. Frank Pasquale, The Black Box Society: The Secret Algorithms That Control Money and Information 3–8 (2015).

10. Gary Edmond et al., Admissibility Compared: The Reception of Incriminating Expert Evidence, 16 Int’l J. Evidence & Proof 1, 7–10 (2012).

11. Brandon L. Garrett, Convicting the Innocent: Where Criminal Prosecutions Go Wrong 45–51 (2011).

12. Maneka Gandhi v. Union of India, (1978) 1 SCC 248, 279–281.

13. Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 589–595 (1993); R v. Dlugosz, [2013] EWCA Crim 2.

14. A. Kumar et al., Artificial Intelligence in the Identification of the Deceased: A Narrative Review with Implications in Forensic Science, 27 Behav. Sci. & L. 1, 2–4 (2024).

15. A. Sharma et al., A Narrative Review on Application of Artificial Intelligence in Forensic Sciences, J. Indian Oral Health (2025).

16. Revolutionizing Forensic Science: The Role of Artificial Intelligence in Evidence Analysis, Int’l J. Interdisciplinary Approaches Psych. (2025).

17. Vaibhavi Tiwari, The Future of Artificial Intelligence in Forensics: Advancements, Challenges, and Ethical Considerations 3–6 (2024).

18. Artificial Intelligence in Forensic Science, IGI Global (2023).

19. Digital Detectives: Exploring the Integration of Artificial Intelligence in Indian Forensic Medicine Praveen Dixit, Digital Detectives: Exploring the Integration of Artificial Intelligence in Indian Forensic Medicine (2024).

20. Bureau of Police Research & Development, Use of Artificial Intelligence in Aiding Forensic & Legal Enforcement (Indian Police Journal, Jan–June 2025).

21. e-Committee, Supreme Court of India, Digital Courts: Vision & Roadmap, Phase III (2022).

22. Press Info. Bureau, Gov’t of India, Press Release on Use of Artificial Intelligence in Judiciary and Criminal Justice System (2024).

23. INTERPOL, Disaster Victim Identification Guide 45–52 (2018).

24. A. Ross & K. Jain, Postmortem Biometric Identification, 43 IEEE Trans. Pattern Analysis & Mach. Intelligence 1, 3–5 (2021).

25. INTERPOL, Disaster Victim Identification Guide 45–52 (2018).

26. State of H.P. v. Jai Lal, (1999) 7 SCC 280, 286–287.

27. A. Kumar et al., Artificial Intelligence in the Identification of the Deceased: A Narrative Review with Implications in Forensic Science, 27 Behav. Sci. & L. 1, 2–4 (2024).

28. A. Ross & K. Jain, Postmortem Biometric Identification, 43 IEEE Trans. Pattern Analysis & Mach. Intelligence 1, 3–5 (2021).

29. Revolutionizing Forensic Science: The Role of Artificial Intelligence in Evidence Analysis, Int’l J. Interdisciplinary Approaches Psych. (2025).

30. Vaibhavi Tiwari, The Future of Artificial Intelligence in Forensics: Advancements, Challenges, and Ethical Considerations 14–18 (2024).

31. Ibid.

32. State of H.P. v. Jai Lal, (1999) 7 SCC 280, 286–287.

33. Ramesh Chandra Agrawal v. Regency Hosp. Ltd., (2009) 9 SCC 709, 719–721.

34. Aditya Agrawal, The Admissibility and Weight of Expert Opinion Under the Indian Evidence Act: A Legal Perspective, 16(2) IJSAT–Int’l J. Sci. & Tech.(2025).

35. Ibid.

36. Vaibhavi Tiwari, The Future of Artificial Intelligence in Forensics: Advancements, Challenges, and Ethical Considerations 14–18 (2024).

37. Artificial Intelligence in Forensic Science, IGI Global (2023).

38. State of H.P. v. Jai Lal, (1999) 7 SCC 280, 286–287.

39. Vaibhavi Tiwari, The Future of Artificial Intelligence in Forensics: Advancements, Challenges, and Ethical Considerations 14–18 (2024).

40. INDIA CONST. art. 21.

41. Maneka Gandhi v. Union of India, (1978) 1 SCC 248, 279–281.

42. Selvi v. State of Karnataka, (2010) 7 SCC 263, 319–323.

43. Fatbardha Ismaili & Shkelqim Sulejmani, Human Rights and Criminal Procedure, 12 Justicia Int’l J. Legal Sci. 364, 364–68 (2024).

44. Ibid.

45. Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579 (1993).

46. Brandon L. Garrett, Autopsy of a Crime Lab: Exposing the Flaws in Forensics 87–93 (2021).

47. R v. Dlugosz, [2013] EWCA Crim 2, ¶¶ 16–21.

48. Suzanne Bell, Forensic Science: An Introduction to Scientific and Investigative Techniques 112–15 (CRC Press 2025).

49. Gary Edmond, Judicial Gatekeeping and the Limits of Science, 36 U. New S. Wales L.J. 863, 870–75 (2013).

50. David L. Faigman et al., Modern Scientific Evidence: The Law and Science of Expert Testimony 1–12 (2023).

51. Law Commission of England and Wales, Expert Evidence in Criminal Proceedings in England and Wales 23–29 (Law Com No. 325, 2011).

52. Vaibhavi Tiwari, The Future of Artificial Intelligence in Forensics: Advancements, Challenges, and Ethical Considerations 14–18 (2024).

53. Bureau of Police Research & Development, Use of Artificial Intelligence in Aiding Forensic & Legal Enforcement (Indian Police Journal, Jan–June 2025).

54. Malay Kumar Ganguly v. Dr. Sukumar Mukherjee, (2009) 9 SCC 221, 252–254.

55. Law Commission of England and Wales, Expert Evidence in Criminal Proceedings in England and Wales 23–29 (Law Com No. 325, 2011).

56. Brandon L. Garrett, Autopsy of a Crime Lab: Exposing the Flaws in Forensics 87–93 (2021).

57. INTERPOL, Guidance on the Use of AI in Law Enforcement (2021).

58. National Institute of Standards and Technology (NIST), AI Risk Management Framework (2023).

59. e-Committee, Supreme Court of India, Digital Courts: Vision & Roadmap, Phase III (2022).

60. Justice K.S. Puttaswamy (Retd.) v. Union of India, (2017) 10 SCC 1, 319–321.