Authored By: Refilwe Moraba
University of South Africa
Introduction
The rapid advancement of Artificial Intelligence (AI) has presented new evidentiary challenges. There is now highly convincing synthetic media — often referred to as “deepfakes” — which can imitate real individuals. With new ways to fabricate realistic visual and audio content with alarming accuracy, can courts still rely on digital evidence?
Artificial Intelligence was created for personalised experiences, swift data analysis, reduced human error, and the automation of repetitive tasks, among many other legitimate purposes. However, there are those who use AI for malicious purposes, including the creation of synthetic media that can mimic real people, creating difficulties for the law of evidence. Courts have long relied on digital evidence — including videos and audio recordings — in legal proceedings to ensure fair trials. Synthetic media is now raising serious concerns about the authenticity and reliability of such evidence. Genuine evidence might be obscured, while in some cases synthetic media might lead to wrongful convictions.
The rise of AI has introduced evidentiary challenges that expose gaps in the existing legal framework, requiring urgent verification measures and reforms suited to this new era.
This article examines the rising sophistication of AI and the potential shortcomings in the credibility of digital evidence, and further considers whether legal reforms are necessary or whether existing laws are sufficient.
Legal Framework
South African law does not govern synthetic media specifically. However, the courts have consistently emphasised the importance of credibility and reliability in the admission of evidence.
Every individual has the right to a fair trial and to challenge evidence, provided it is reliable. The Constitution of the Republic of South Africa, 19961 — specifically section 35(3) — ensures those rights for every accused person. Through this provision, individuals may challenge digital evidence and require that it be proven unmanipulated and that it is what it purports to be.
The Law of Evidence Amendment Act 45 of 19882 regulates the admissibility of hearsay evidence. This Act was introduced to address traditional hearsay, including written and oral statements, long before an era where AI-generated videos, images, and audio recordings had to be assessed for authenticity. AI-generated media may now fall within the scope of hearsay evidence, depending on the nature of the evidence and whether its admission would be in the interest of justice. This Act therefore allows courts to consider AI-generated media, provided it is authentic and reliable.
Furthermore, the Electronic Communications and Transactions Act 25 of 20023 is the cornerstone statutory provision in this area. Section 15 of the Act establishes that data messages are to be treated as traditional evidence if proven authentic, and clarifies that digital evidence must not be denied admissibility solely because of its electronic form, as it carries legal validity. However, the Act does provide that the reliability of such evidence and the manner in which it was generated must be fully evaluated.
These legal frameworks provide the foundation for admissibility, but with synthetic media there is a real risk of misinterpretation and manipulation of digital evidence. This raises the question of whether existing frameworks can adequately address new and advanced technologies.
Legislative Analysis
S v Ndiki and Others [2007] 2 All SA 185 (Ck)4 is a landmark South African case which established that digital evidence is admissible if proven reliable. It is a significant case in that it confirmed computer-generated and electronic evidence could be admitted, and that such evidence cannot be rendered automatically inadmissible simply because it depends on a computer system — its authenticity must first be evaluated.
The ruling in this case aligned with the Law of Evidence Amendment Act 45 of 1988, which permits hearsay evidence where admission is in the interest of justice. The court’s approach to electronic messages — treating them as legitimate evidence rather than dismissing them on grounds of form — was also consistent with the Electronic Communications and Transactions Act 25 of 2002. The decision demonstrated the law’s capacity to adapt to modern technology by accepting digital evidence.
The judgment in S v Ndiki illustrates that the South African evidentiary approach is a hybrid one, drawing on traditional legal frameworks alongside emerging technologies. The case also exposed the risks associated with those technologies, underscoring the crucial need for safeguards to guarantee evidential integrity.
Critical Analysis and Argument
AI is advancing daily and is increasingly accessible to the public, making the creation of fabricated content easier than ever before. This poses a major evidentiary risk: courts may admit highly convincing fabricated content that appears reliable and credible, resulting in judicial proceedings being undermined. This article argues that although the law of evidence provides a basic framework for assessing the authenticity and admissibility of evidence, South Africa may remain ill-equipped to face this new era of sophisticated synthetic media. While digital media has historically been regarded as inherently credible and reliable, the admissibility of AI-generated content on the same basis is increasingly untenable.
At present, courts are required to assess, detect, and challenge sophisticated digital evidence — a task that is unrealistic without technical expertise and the appropriate technological tools. This difficulty is further compounded by institutional limitations. The resulting gap between current legal standards and technological reality represents a genuine threat to the law of evidence. Without reform, digital evidence risks becoming progressively less reliable as a category. The current legal framework needs to evolve, as it may prove insufficient to address the emerging risks posed by AI.
While technology continues to advance, the law must evolve in step, introducing stricter standards governing the admissibility of digital evidence. Courts must rely on expert evidence while advancing authentication standards. Without such developments, the risk of errors will grow, deterring fairness and justice, and ultimately eroding public confidence in the administration of fair trials.
Under the existing law of evidence, digital media has generally been presumed reliable unless proven otherwise. However, with the rise of highly convincing fabricated content, sustaining that presumption is increasingly difficult. This article argues that South African courts can continue to rely on digital evidence — but only provided such evidence is rigorously scrutinised and properly authenticated through expert testimony. Increased caution will ensure the integrity of proceedings and the fairness of outcomes.
Conclusion
This article’s argument is not solely whether courts can still rely on digital evidence, but rather the manner in which such reliance should operate — which must be conditional on proper authentication. The analysis above indicates that the reliability of digital evidence is not guaranteed, regardless of its admissibility in law. The emergence of accurate digital fabrication is challenging traditional safeguards and threatens to undermine evidential integrity.
The analysis suggests that South African courts are currently relying on digital evidence through flexible statutory provisions, demonstrating a willingness to acknowledge new technological developments in modern litigation without necessarily introducing new legal frameworks. However, this flexibility must be matched with rigour.
It is apparent that, within a context-sensitive legal framework that acknowledges both the evidentiary value and the potential limitations of digital evidence, South African courts can continue to rely on such evidence. However, that reliance must be exercised with caution and must not be absolute, in order to avoid the risks of manipulation.
Given the sophistication of synthetic media, mandatory technical expertise in judicial proceedings has become essential. Without replacing the current framework, a critical step forward would be the introduction of statutory guidelines governing the evaluation and authentication of digital evidence. Such developments would not displace existing frameworks but would instead strengthen safeguards to ensure that digital evidence remains credible and reliable within an evolving technological reality.
Reference(S):
Cases
- S v Ndiki and Others [2007] 2 All SA 185 (Ck).
Legislation
- Constitution of the Republic of South Africa, 1996.
- Law of Evidence Amendment Act 45 of 1988.
- Electronic Communications and Transactions Act 25 of 2002.
Secondary Sources
- Swales L “An Analysis of the Regulatory Environment Governing Hearsay Electronic Evidence in South Africa: Suggestions for Reform – Part One” PER / PELJ 2018(21).
Footnote(S):
1 Constitution of the Republic of South Africa, 1996, s 35(3).
2 Law of Evidence Amendment Act 45 of 1988.
3 Electronic Communications and Transactions Act 25 of 2002, s 15.
4 S v Ndiki and Others [2007] 2 All SA 185 (Ck).
