Role of generative artificial intelligence in crime investigation in South Africa

Abstract

The integration of generative artificial intelligence (GenAI) technologies into crime investigation represents a paradigm shift in modern law enforcement capabilities.  This literature review examines the role, applications, opportunities, and challanges of implementing generative AI systems, including Large Language Models (LLMs), Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs),  and diffusion models within the context of South African crime investigation.  Through systemic analysis of current research, this article explores how GenAI can enhance forensic analysis, evidence processing, suspect identification, and investigation decision-making while addressing critical concerns regarding bias, privacy, legal admissibility, and ethical implications.  The review reveals that while generative AI offers transformative potential for addressing South Africa's unique crime challenges, successful implementation requires careful governance frameworks, staged deployment strategies, legal alignment, and capacity building.  Key findings indicate that retrieval-augmented generation systems, forensic image synthesis, and synthetic data generation present immediate opportunities but must be balanced against documented risks, including algorithmic fabrication, false memory induction, and systemic bias. This article concludes with evidence-based recommendations for South African law enforcement agencies seeking to responsibly harness generative AI capabilities. 

Keywords: {'name': 'Generative Artificial Intelligence, Crime Investigation, Law Enforcement, South Africa, Machine Learning, Forensic Science, Large Language Models, Digital Forensics'}

Authors & affiliations

Lehlohonolo Wonderboy Mahlatsi
University of South Africa

Deposited references

Supplied references, not a list of publications citing this article.

DOI: 10.55041/IJSREM34747

DOI: 10.21533/pen.v8.i3.1168

Constitutional and legal frameworks for AI in South African criminal justice. (2024). South African Law Journal, 141(2), 312–345.

Conversational AI powered by large language models amplifies false memories in witness interviews. (2024). arXiv Preprint. http://arxiv.org/abs/2408.04681v1

DOI: 10.1371/journal.pone.0332802

DOI: 10.55248/gengpi.5.1024.2904

Faqir, R. S. A. (2021). Digital criminal investigations in the era of artificial intelligence: A comprehensive overview. Zenodo. https://doi.org/10.5281/zenodo.4766706

DOI: 10.1093/police/paaf027

DOI: 10.25136/2409-7810.2024.1.68894

DOI: 10.1016/j.fsidi.2025.301939

DOI: 10.1016/j.aiopen.2024.09.002

DOI: 10.1016/j.joitmc.2023.100151

Mabokela, K. R., & Schlippe, T. (2022). A sentiment corpus for South African under-resourced languages in a multilingual context. In Proceedings of the 1st Annual Meeting of the ELRA/ISCA Special Interest Group on Under-Resourced Languages (SIGUL 2022) (pp. 70–77). European Language Resources Association. https://aclanthology.org/2022.sigul-1.9

DOI: 10.1515/jisys-2022-0223

DOI: 10.1109/EEITE61750.2024.10654427

DOI: 10.37502/IJSMR.2025.8515

DOI: 10.4018/979-8-3373-0543-1.ch013

DOI: 10.1177/03400352251348243

DOI: 10.11591/ijai.v14.i3.pp1743-1751

DOI: 10.5121/ijaia.2021.12106

South African crime statistics and law enforcement challenges. (2023). South African Journal of Criminal Justice, 36(2), 201–225.

DOI: 10.3390/app15137105

DOI: 10.1007/978-3-031-22552-9_12