VidhiAagaz
International Journal ofLaw Management & Humanities
ISSN 2581-5369 · Open access
International Journal ofLaw Management & Humanities
Open accessArticlesVol. 9 No. VI (2026)pp. 1223–1232

Use of AI in Criminal Investigations Under BNSS 2023

KG
Kaviyasri GStudent at Anvikshiki School of Law, Sri Chandrasekharendra Saraswathi Viswa Mahavidyalaya, Kanchipuram, Tamil Nadu, India
0views0downloads
Published 20 November 2026 · pp. 1223–1232 · Vol. 9 No. VI (2026)
Abstract

Artificial Intelligence (AI) is transforming criminal investigation and forensic science by enabling the rapid analysis of large volumes of digital and physical evidence. The integration of AI into criminal investigations enhances the efficiency of digital forensics, video and image analysis, biometric identification, DNA analysis, ballistic examination and cybercrime detection. Machine learning, computer vision and natural language processing assist investigators in identifying patterns, detecting anomalies, correlating evidence and reducing the time required for manual examination. Compared with traditional forensic methods, AI-driven systems offer greater scalability and automated evidence processing, although their accuracy and reliability depend on data quality, algorithmic design and expert validation. In India, provisions of the Bharatiya Nagarik Suraksha Sanhita, 2023 (BNSS), particularly Sections 173, 176(3) and 185, provide a relevant procedural framework for electronic reporting, forensic crime-scene examination and the audio-video recording of searches. However, the adoption of AI in criminal investigations raises significant concerns regarding algorithmic bias, privacy, transparency, evidentiary reliability and legal admissibility. This study examines the major applications, benefits, limitations and legal implications of AI in criminal investigation, emphasising the importance of human oversight, scientific validation and procedural safeguards. It concludes that AI should function as an assistive investigative technology rather than a replacement for professional judgment and established forensic procedures.

Keywords
Artificial IntelligenceCriminal InvestigationDigital ForensicsForensic ScienceMachine LearningBNSSEvidence AnalysisCybercrime
Contents
Full text

Introduction

Artificial Intelligence (AI) is a modern technology that enables machines and computer systems to perform tasks that require human intelligence, such as learning, reasoning, problem-solving and decision-making; AI applications are specialised in particular tasks and, in some cases, can surpass human capabilities.1 AI has become an important tool in criminal investigation and forensic science because of the increasing volume of evidence, particularly digital evidence. Traditional investigative methods often require considerable time and human effort to analyse such evidence, whereas AI helps process large amounts of information quickly and efficiently.2

AI is used in digital forensics, facial recognition, CCTV analysis, cybercrime detection, DNA examination and ballistic analysis. Machine learning and computer vision help identify patterns, detect suspicious activities and connect evidence from different sources. Natural language processing also assists in analysing emails, text messages, documents and witness statements.3 These applications help investigators save time, reduce manual effort and identify important clues.

In India, the Bharatiya Nagarik Suraksha Sanhita, 2023 (BNSS) provides a relevant legal framework for technology-assisted criminal investigation. Sections 173, 176(3) and 185 address electronic reporting, forensic examination of crime scenes in specified cases and audio-video recording of police searches, respectively.4 However, the use of AI also raises concerns about privacy, algorithmic bias, accuracy and the admissibility of evidence in court.

Therefore, AI should be used as an assisting tool rather than a replacement for human judgment. This study examines the applications, advantages, challenges and legal implications of AI in criminal investigation, emphasising the importance of scientific validation, transparency and human supervision to ensure fair and reliable outcomes.

Review of literature

Ortega Anderez et al. (2021) examine the role of emerging technologies in crime prevention and control.5 Their research highlights the use of GPS tracking, video surveillance, the Internet of Things, machine learning and edge computing in identifying and preventing criminal activity. The authors observe that technological tools can improve the ability of law enforcement agencies to detect criminal behaviour and respond to incidents. At the same time, they stress the importance of addressing limitations relating to privacy, accuracy, affordability and the misuse of technology.

Pawar, Taneja and Singh (2017), in their study Use of Technology in Criminal Investigation by Police, analyse the impact of technological tools on conventional policing.6 The authors discuss the use of mobile data terminals, GPS, drones and smartphones in police investigations. These technologies provide real-time information and improve communication. The study argues that technology can substantially improve the speed and effectiveness of police investigation, particularly in surveillance, location tracking and coordination between investigating officers.

Yadav (2026), in the study Overhauling India’s Criminal Law Framework: A Doctrinal Comparison of BNS 2023 with IPC 1860 and its Interface with BNSS and BSA, examined the major structural and substantive changes introduced by the Bharatiya Nyaya Sanhita, 2023 (BNS) in comparison with the Indian Penal Code, 1860 (IPC).7 Using doctrinal and comparative approaches, the study highlighted significant reforms, including the introduction of provisions relating to organised crime and terrorism, the replacement of the traditional sedition provision with Section 152, and greater emphasis on offences against women and children. The study also examined the relationship between the BNS, the Bharatiya Nagarik Suraksha Sanhita (BNSS) and the Bharatiya Sakshya Adhiniyam (BSA), particularly with regard to digital procedures, forensic investigation and electronic evidence. Yadav concluded that although the new criminal-law framework modernises substantive and procedural criminal justice, broad national-security provisions and overlaps with special laws may create concerns regarding discretionary enforcement. The study therefore emphasised the need for constitutional safeguards, institutional capacity and proper interpretation to ensure that criminal-law reforms remain consistent with fundamental rights.

In State (NCT of Delhi) v. Navjot Sandhu, the Supreme Court adopted a relatively flexible approach towards the admissibility of electronic records, holding that secondary evidence of call records could be led under Sections 63 and 65 of the Evidence Act irrespective of compliance with Section 65B.8 In Anvar P.V. v. P.K. Basheer, the Supreme Court overruled that view and clarified the importance of compliance with the statutory requirements governing electronic evidence under the then-existing Evidence Act: an electronic record offered as secondary evidence is not admissible unless the requirements of Section 65B are satisfied.9 The corresponding provision is now Section 63 of the Bharatiya Sakshya Adhiniyam, 2023.10 In Vikram Singh v. State of Punjab, the Supreme Court held that a tape-recorded ransom call produced on the original cassette was primary evidence, so that no certificate under Section 65B was required for its admission; the certificate is mandatory only where an electronic record is offered as secondary evidence.11 These decisions illustrate the judiciary’s attempt to balance technological developments with evidentiary reliability.

AI and traditional digital forensics

Traditional forensic examination is slow, whereas AI-based automated evidence processing promises to expedite analysis while increasing case-processing capacity.12 In the traditional method, the examiner checks material one item at a time; AI-based triage could direct the analysis first to the devices most likely to contain information that advances the case.13 In the traditional method, accuracy also depends on the person who is checking: reviewing a huge volume of images and videos by hand is time-consuming and carries the potential for human error due to fatigue.14 Finally, in the traditional method, evidence has to be linked manually, whereas automated techniques are being developed to analyse the very large number of timestamps on a device and infer a usable history of activity from them.15

Legal framework

Section 173 of the Bharatiya Nagarik Suraksha Sanhita, 2023 (BNSS) provides the procedure for recording information relating to a cognizable offence. Information about a cognizable offence can be given orally or by electronic communication, irrespective of the area where the offence was committed. If the information is given by electronic communication, it is taken on record on being signed within three days by the person giving it. The provision also contains special safeguards for recording information relating to specified offences against women and persons with disabilities.16 Section 173 is significant to AI-based criminal investigation as it legally recognises electronic communication at the initial stage of criminal investigation, thereby facilitating the integration of digital technologies into the reporting and investigation process.

Section 176(3) of the BNSS requires the officer in charge of a police station, on receiving information relating to an offence punishable with imprisonment for seven years or more, to cause a forensic expert to visit the crime scene to collect forensic evidence and to cause the process to be videographed on a mobile phone or any other electronic device, from a date to be notified by the State Government within five years.17 This provision is significant for AI-based criminal investigation as it promotes scientific and digital methods of evidence collection, creating scope for AI-assisted forensic analysis and crime-scene examination.

Section 185 of the BNSS empowers the officer in charge of a police station, or a police officer making an investigation, to search a place within the limits of the police station when there are reasonable grounds to believe that anything necessary for the investigation may be found there and cannot otherwise be obtained without undue delay. The provision also requires the search to be recorded through audio-video electronic means, preferably by mobile phone.18 Section 185 is relevant to AI-assisted criminal investigation because it facilitates the use of digital recording and technological tools during searches, improving transparency, documentation and preservation of evidence.

Applications

In digital forensics, AI helps investigators examine large amounts of digital information held on phones, computers and other devices, including encrypted data, and retrieve useful information connected to a crime. In cybercrime investigation, it is used to analyse network traffic and to detect malware.19

AI uses computer vision to examine CCTV footage, body-worn camera videos and photographs. Such systems can detect and recognise faces, identify people, track objects, identify vehicles and recognise licence plates,20 and researchers are developing algorithms to identify people, cars and weapons in video without human intervention.21 Facial recognition identifies persons of interest by comparing the features of a face in an image or video with faces held in a database,22 and such comparisons should be made only against legally authorised databases. AI voice profiling can also support the identification of possible suspects from voice recordings.23

Natural language processing can create automated transcripts of recorded conversations and translate between languages, and it can process large volumes of written material, such as emails, online chats and documents, to extract relevant information, saving investigators time on manual review.24

In predictive policing, AI examines previous crime records, the time and place of earlier incidents and other data to predict where and when crimes are likely to occur. This helps police plan patrol routes and allocate resources to areas that may require greater attention, although such predictions are not certain.25

Major challenges

The first challenge is algorithmic bias. Where the data used by an AI system is the product of flawed or biased past policing, the system may incorporate those biases,26 which can result in certain communities being unnecessarily targeted or subjected to increased surveillance. Second, extensive surveillance using AI may affect people’s privacy, freedom and the protection of their personal information; the Supreme Court has recognised privacy as a fundamental right protected by Article 21 of the Constitution.27 Third, the lack of transparency in AI systems, often described as the “black box” problem, makes it difficult for courts and defence lawyers to understand28 and verify how the AI arrived at a particular result.29

How AI handles DNA evidence

AI helps investigators analyse DNA samples that are mixed with DNA from different people, present in very small amounts or degraded.30 Probabilistic genotyping software applies algorithms to interpret such complex mixtures.31 AI may also improve electropherogram systems by reducing or eliminating artefacts, so that genuine genetic signals are distinguished from noise.32 Biological evidence can further be used to predict certain physical features, such as pigmentation traits,33 and AI may help estimate a person’s age from crime-scene biological material.34

How AI handles ballistics evidence

AI-based image analysis examines the tiny marks on bullets and cartridge cases to help identify similarities between them.35 Automated systems also compare images of cartridge cases with records stored in ballistic databases to identify possible links to firearms.36

Limitations and oversight

When AI systems do not clearly explain how they reach their conclusions, it can create legal and constitutional difficulties in determining whether their results can be accepted as evidence in court.37 AI is used to assist experts in examining and sorting evidence, but the final analysis and decision should remain under human supervision rather than relying entirely on AI.

Core potential

The core potential of AI lies in speed and efficiency: it helps investigators analyse large amounts of complex evidence, such as digital files, CCTV footage, mobile phone data and DNA information, and find useful clues and patterns in less time than manual examination.38 It can detect objects such as weapons in video footage and can make examination more consistent by reducing mistakes that occur through fatigue.39 However, human experts must still review and interpret the results to ensure that the findings are accurate and reliable.

Critical analysis

Artificial Intelligence has the potential to transform criminal investigation by making the collection, examination and correlation of evidence faster and more systematic. Provisions such as Sections 173, 176(3) and 185 of the BNSS create an important legal foundation for the use of electronic communication, forensic expertise, videography and audio-video recording during investigation. However, the BNSS does not expressly provide a legal framework governing the use of AI itself in criminal investigations. The major advantage of AI is its ability to process enormous quantities of digital, forensic, biometric and video data within a short period. It can assist investigators in identifying patterns that may not be easily detected through traditional methods. AI can therefore function as an effective investigative support tool in areas such as digital forensics, facial recognition, DNA analysis, ballistics and cybercrime investigation.

At the same time, excessive reliance on AI creates serious legal and ethical concerns. AI systems are trained using existing datasets, and biased or incomplete datasets may produce discriminatory results.40 Facial-recognition systems, predictive policing and automated suspect identification may therefore result in false positives and unnecessary targeting of individuals.41 Privacy is another major concern because criminal investigations increasingly involve the collection and analysis of large quantities of personal data.42

The black-box nature of some AI systems creates an additional difficulty. If investigators or courts cannot understand how an AI system reached a particular conclusion, it becomes difficult for the accused to challenge that evidence effectively.43 This raises concerns relating to fair trial, natural justice, reliability and the accused’s right to defend themselves. AI-generated results should therefore not automatically be treated as conclusive evidence.

Another important issue is human accountability. AI can identify patterns and assist forensic experts, but it cannot replace the legal responsibility of investigators, forensic experts, prosecutors and courts. AI outputs should be independently verified before being relied upon for arrest, charge-sheeting or conviction. Proper documentation of the AI system, the data used, the methodology, the accuracy rate and the human verification process is essential for maintaining the integrity of criminal investigations.

Therefore, AI should be viewed as an assistive investigative technology rather than an autonomous decision-maker. Its effectiveness depends not only on technological accuracy but also on transparency, human oversight, procedural safeguards and compliance with constitutional rights.

Suggestions

•  Develop specific AI guidelines: clear guidelines should be formulated for the use of AI in criminal investigations, including permissible uses, limitations and accountability mechanisms.

•  Mandatory human verification: AI-generated findings should be examined and verified by qualified forensic experts or investigators before being relied upon.

•  Maintain an audit trail: investigators should record the AI tool used, the data analysed, the methodology followed, the results obtained and the human verification carried out.

•  Protect privacy: AI investigations involving facial recognition, biometric information, mobile data and cloud data should follow appropriate privacy and data-protection safeguards.

•  Address algorithmic bias: AI systems used by law-enforcement agencies should undergo regular testing for discriminatory outcomes and accuracy across different populations.

•  Ensure transparency: where AI-generated evidence is relied upon in court, the defence should have a meaningful opportunity to challenge its reliability and methodology, subject to legitimate security and privacy restrictions.

•  Training of investigators: police officers and forensic personnel should receive specialised training in AI, digital forensics, cybercrime and the limitations of automated systems.

•  Accreditation and validation: AI forensic tools should undergo scientific validation and quality testing before being routinely used in criminal investigations.

•  Preserve original evidence: AI analysis should never replace preservation of the original digital or physical evidence.

•  Independent oversight: an appropriate regulatory or institutional mechanism should periodically review the use of AI by law-enforcement agencies to ensure legality, accuracy and accountability.

Conclusion

Artificial Intelligence is emerging as an important tool in modern criminal investigation. Its ability to process large volumes of digital information, identify patterns and analyse forensic evidence can significantly improve the speed and efficiency of investigations. The BNSS, particularly Sections 173, 176(3) and 185, recognises the growing importance of electronic communication, forensic examination and audio-video documentation in the investigative process. However, technological advancement must remain subject to law. Algorithmic bias, privacy violations, lack of transparency, inaccurate predictions and overdependence on automated systems may adversely affect individual rights and the fairness of criminal proceedings. AI-generated results should therefore be treated as investigative assistance and not as an unquestionable substitute for human expertise.

A balanced approach is necessary in which AI is used to strengthen scientific investigation while investigators, forensic experts and courts retain ultimate responsibility for evaluating evidence. With proper regulation, transparency, human verification, privacy safeguards and continuous technological validation, AI can become a powerful instrument for achieving efficient, accurate and fair criminal investigations under the BNSS, 2023.

*****

Footnotes

1. INTERPOL & UNICRI, Towards Responsible AI Innovation: Second INTERPOL-UNICRI Report on Artificial Intelligence for Law Enforcement 51 (2020).

2. Xiaoyu Du et al., SoK: Exploring the State of the Art and the Future Potential of Artificial Intelligence in Digital Forensic Investigation, in Proceedings of the 15th International Conference on Availability, Reliability and Security (2020), https://arxiv.org/abs/2012.01987.

3. INTERPOL & UNICRI, supra note 1, at 17, 20–21; Christopher Rigano, Using Artificial Intelligence to Address Criminal Justice Needs, NIJ J., no. 280, Dec. 2018, https://nij.ojp.gov/topics/articles/using-artificial-intelligence-address-criminal-justice-needs; President’s Council of Advisors on Science and Technology, Forensic Science in Criminal Courts: Ensuring Scientific Validity of Feature-Comparison Methods 114 (2016).

4. The Bharatiya Nagarik Suraksha Sanhita, 2023, No. 46, Acts of Parliament, 2023, §§ 173, 176(3), 185 (India).

5. Dario Ortega Anderez et al., The Rise of Technology in Crime Prevention: Opportunities, Challenges and Practitioners Perspectives §§ 2, 4 (Jan. 26, 2021) (unpublished manuscript), https://arxiv.org/abs/2102.04204.

6. Leena Pawar, Anita Taneja & Bhim Singh, Use of Technology in Criminal Investigation by Police, 3 J. Legal Stud. & Rsch. 68 (2017), https://thelawbrigade.com/wp-content/uploads/2019/05/Leena-Bhim-Anita.pdf.

7. Chandra Kishore Yadav, Overhauling India’s Criminal Law Framework: A Doctrinal Comparison of BNS 2023 with IPC 1860 and its Interface with BNSS and BSA, 2 J. Advanced Interdisc. Stud. 13 (2026), https://journals.eduacademic.co.in/index.php/jais/article/view/324.

8. State (NCT of Delhi) v. Navjot Sandhu, (2005) 11 SCC 600, ¶ 150 (India).

9. Anvar P.V. v. P.K. Basheer, (2014) 10 SCC 473, ¶ 22 (India).

10. The Bharatiya Sakshya Adhiniyam, 2023, No. 47, Acts of Parliament, 2023, § 63 (India).

11. Vikram Singh v. State of Punjab, (2017) 8 SCC 518 (India).

12. Du et al., supra note 2.

13. Du et al., supra note 2, § 3.2.

14. Rigano, supra note 3.

15. Du et al., supra note 2, § 3.5.

16. Bharatiya Nagarik Suraksha Sanhita, supra note 4, § 173.

17. Bharatiya Nagarik Suraksha Sanhita, supra note 4, § 176(3).

18. Bharatiya Nagarik Suraksha Sanhita, supra note 4, § 185.

19. Du et al., supra note 2, §§ 3.1, 3.3, 3.4.

20. INTERPOL & UNICRI, supra note 1, at 17.

21. Rigano, supra note 3.

22. INTERPOL & UNICRI, supra note 1, at 51.

23. INTERPOL & UNICRI, supra note 1, at 14 (noting that voice profiling has “some distance to go before it becomes viable in court”).

24. INTERPOL & UNICRI, supra note 1, at 20–21.

25. INTERPOL & UNICRI, supra note 1, at 30; Rigano, supra note 3.

26. Aziz Z. Huq, Constitutional Rights in the Machine-Learning State, 105 Cornell L. Rev. 1875, 1888 (2020).

27. Justice K.S. Puttaswamy (Retd.) v. Union of India, (2017) 10 SCC 1 (India).

28. INTERPOL & UNICRI, supra note 1, at 34.

29. See Rebecca Wexler, Life, Liberty, and Trade Secrets: Intellectual Property in the Criminal Justice System, 70 Stan. L. Rev. 1343, 1343 (2018) (developers often claim that details of how their tools work are trade secrets and refuse to disclose them to criminal defendants or their attorneys).

30. Rigano, supra note 3.

31. President’s Council of Advisors on Science and Technology, supra note 3, at 78–79 (noting that such programs “still require careful scrutiny” of their scientific validity).

32. Ioannis Ketsekioulafis et al., Artificial Intelligence in Forensic Sciences: A Systematic Review of Past and Current Applications and Future Perspectives, 16 Cureus e70363 (2024), https://doi.org/10.7759/cureus.70363.

33. Manfred Kayser, Forensic DNA Phenotyping: Predicting Human Appearance from Crime Scene Material for Investigative Purposes, 18 Forensic Sci. Int’l: Genetics 33 (2015), https://doi.org/10.1016/j.fsigen.2015.02.003.

34. Ketsekioulafis et al., supra note 32.

35. President’s Council of Advisors on Science and Technology, supra note 3, at 114 (describing an algorithm that extracts a quantitative signature from a three-dimensional image of a bullet, compares it across samples and produces a “matching score”).

36. Id. (noting that the United States National Integrated Ballistic Information Network database “uses older (non-3D) technology and does not provide a scoring function or confidence assignment to each candidate match”).

37. INTERPOL & UNICRI, supra note 1, at 34.

38. Du et al., supra note 2; Rigano, supra note 3.

39. Rigano, supra note 3.

40. Huq, supra note 26, at 1888.

41. Patrick Grother, Mei Ngan & Kayee Hanaoka, Nat’l Inst. of Standards & Tech., NISTIR 8280, Face Recognition Vendor Test (FRVT) Part 3: Demographic Effects 2 (2019), https://doi.org/10.6028/NIST.IR.8280 (finding that, across demographic groups, “false positives rates often vary by factors of 10 to beyond 100 times”).

42. Puttaswamy, (2017) 10 SCC 1 (India).

43. See Wexler, supra note 29, at 1343.

How to cite

Kaviyasri G, Use of AI in Criminal Investigations Under BNSS 2023, 9 Int'l J.L. Mgmt. & Human. 1223 (2026), https://doi.org/10.63108/IJLMH.13176.

Cite this article

Kaviyasri G, Use of AI in Criminal Investigations Under BNSS 2023, 9 Int'l J.L. Mgmt. & Human. 1223 (2026), https://doi.org/10.63108/IJLMH.13176.
Chat with us