Lucknow, October 23: Artificial intelligence (AI) can play a significant role in making forensic science and criminal investigations faster, more effective and more accurate. From fingerprint and DNA analysis to the examination of CCTV footage, digital evidence and the identification of links between individuals and incidents associated with a crime, AI can enhance the capabilities of investigative agencies. However, treating AI-generated findings as conclusive without verification can pose risks to investigations and judicial proceedings.
These observations were made by experts during a panel discussion on Artificial Intelligence in Forensics and Investigation. They emphasised that AI is a useful supporting technology in criminal investigations, but cannot replace expert knowledge, human judgement or independent verification of evidence. The discussion was held on Friday as part of three-day ‘TechSangam’ at the Uttar Pradesh State Institute of Forensic Science (UPSIFS) International Dialogue-26, with experts from India and abroad participating.
At the beginning of the panel discussion, moderator DIG Technical Ashish Tiwari welcomed the speakers and participants and highlighted the emerging role of AI in forensic science and criminal investigations.
He underlined the importance of forensic science in crime prevention, detection, investigation and prosecution, while referring to the provisions of the Bharatiya Nyaya Sanhita (BNS) and the role of forensic evidence in the broader criminal justice system.
The panel concluded that AI could become an effective technological tool for investigative agencies, but its utility would depend on reliable data, transparent procedures and human oversight. Alongside the speed offered by technology, maintaining the authenticity of evidence and the credibility of judicial proceedings remains equally important.
During the discussion, Sandeep Joshi said, "AI could accelerate investigations by analysing large volumes of data in a short time. Information that would otherwise take considerable time to examine using conventional methods can be analysed more quickly with AI to identify patterns and potential connections.
This can help investigators obtain crucial information in a timely manner and make better-informed decisions."
He explained that AI could prove useful in comparing fingerprint patterns, analysing complex DNA-related data, identifying suspicious activities in CCTV footage and tracking the movement of individuals across different locations. It could also be used to examine large volumes of forensic data and identify potential links between individuals, incidents, locations and evidence.
Joshi said, "The effectiveness of AI depended on data quality, appropriate software and hardware, and the methodology used for analysis. Therefore, ensuring reliable results requires attention not only to the technology itself but also to the processes through which it is deployed."
The panel also held an extensive discussion on issues relating to the reliability of AI. Experts said that while AI could strengthen investigations, inaccurate or biased data, misleading conclusions and tampering with digital evidence could create serious challenges.
AI may sometimes produce conclusions that are not adequately supported by the available evidence. Similarly, systems trained on biased data could influence investigative outcomes. The risk of misleading evidence being presented through deepfakes and other artificially generated content is also increasing.
Experts stressed the need for independent verification of AI-generated findings, authentication of original evidence and proper documentation of the entire analytical process. Investigators should also cross-check AI-generated results against other reliable sources and conventional forensic methods.
They emphasised that human oversight, accountability, and clear legal and ethical standards must be maintained while using the technology. The discussion also highlighted the principle of ‘AI for security and security of AI’. According to the experts, AI can benefit investigations only when the systems themselves are protected against errors, tampering, misuse and unauthorised access.
Retired Commodore Satish K. Satyapathi said, "Using AI in forensic investigations requires in-depth subject knowledge, practical experience and technical understanding. Investigators must know how to frame the right questions, assess the quality of available data, and identify potential errors or misleading information in AI-generated findings."
He referred to his professional association with Dr G.K. Goswami and recounted an incident related to the 2006 Nithari case in Noida. Recalling a conversation with journalist Karan Singh, he also underscored the importance of guidance, practical experience and acquired expertise in the field of investigation.
Satyapathi said, "AI could analyse information and present potential conclusions, but it could not replace an investigator’s professional judgement. Assessing the reliability of AI-generated findings requires careful attention to the process of evidence collection, the chain of custody, the methodology of analysis and the verification of results."
The admissibility in court of findings generated with the assistance of AI depends on the applicable legal provisions and the reliability of the evidence. Evidence does not become automatically admissible merely because an AI system has produced a finding. The relevant expert must be able to explain the analytical methodology, validate the results, and establish their relevance and reliability.
He emphasised that responsible use of AI in forensic investigations and judicial proceedings requires human oversight, professional expertise and proper verification of evidence.
