In courtrooms across Nigeria, a quiet challenge is emerging. Lawyers are increasingly confronted with AI-Generated evidence they cannot easily verify; screenshots, audio recordings, video clips, and documents that may have been generated, manipulated, or fabricated using artificial intelligence. The statute governing the admissibility of electronic evidence, the Evidence Act 2011 (as amended in 2023), was enacted in an era of emails, PDFs, and CCTV footage. While the Act underwent amendment in 2023, the reforms did not address the emergence of generative AI technologies or the evidentiary issues arising from AI-generated content, including deep fakes, synthetic audio, and large language models.
The question whether Nigeria’s legal infrastructure is adequate to adjudicate cases involving AI-generated evidence is far from hypothetical. While existing legal principles provide some insights for addressing electronic evidence, the current framework was not designed to address the challenges posed by generative AI, and significant gaps remain.
Judicial Foresight and the Recognition of AI-Generated Evidence in Nigeria
Prior to the enactment of the Evidence Act 2011, Nigerian courts had recognised the admissibility of computer-generated evidence. In Esso West Africa Inc. v. T. Oyegbola[1], the Supreme Court acknowledged the realities of technological advancement and the need for the law to adapt to modern methods of record-keeping. The Court stated:
“The law cannot be and is not ignorant of the modern business methods and must not shut its eyes on the mysteries of computer. In modern times, reproduction and inscriptions on ledgers or other documents by mechanical process are commonplace and section 37 cannot therefore only apply to books of account.”
This pronouncement marked an early judicial recognition of computer-generated records as a form of documentary evidence, notwithstanding the absence of a comprehensive statutory framework governing electronic records at the time. The enactment of the Evidence Act 2011 represented a significant legislative development in this regard. For the first time, the admissibility of electronically generated evidence was expressly codified, providing a clear statutory framework for the reception of electronic records in judicial proceedings. Section 84 of the Act consequently became the principal provision governing the admissibility of computer-generated evidence in Nigeria.
What Section 84 of the Evidence Act Actually Says
Section 84(1) of the Evidence Act provides that a statement contained in a document produced by a computer shall be admissible as evidence of any fact, provided the conditions set out in subsection (2) are satisfied. Those conditions require a party to establish that the computer was regularly used during the relevant period, that information of the relevant kind was regularly supplied to it in the ordinary course of activities, that the computer was operating properly throughout, and that the information was derived from information supplied to the computer in the ordinary course of those activities.
To satisfy these conditions, the Act permits a foundation to be laid either through oral evidence or by a certificate of compliance under Section 84(4) of the Evidence Act. The Supreme Court in Kubor & Anor v. Dickson & Ors[2] confirmed that both modes of authentication are available and held that a party seeking to rely on an electronically generated document must do more than merely tender it from the bar. Sufficient proof in line with the requirements of Section 84(2) of the Evidence Act must first be adduced.
The Evidence (Amendment) Act 2023 introduced Sections 84A to 84D into the Evidence Act, expanding the definition of “electronic record” to include data, records, images, and sound stored, received, or transmitted electronically. Section 84A specifically provides that where any law requires information to be in writing or printed form, that requirement is satisfied if the information is made available in electronic form and remains accessible for subsequent reference. These amendments represent a significant step forward in recognising the growing importance of digital records. However, they remain silent on AI-generated content, leaving one of the most transformative technological developments of recent years outside the scope of direct legislative consideration.
The Statutory Gap
The central difficulty lies in the fact that the Evidence Act (As Amended in 2023) does not expressly address artificial intelligence. Section 258(1) of the Act defines a “computer” broadly as any device used for storing and processing information. Some commentators argue that this definition is sufficiently expansive to encompass AI systems, thereby bringing AI-generated outputs within the ambit of Section 84. While that interpretation may be possible, it remains an exercise in statutory construction rather than a reflection of clear legislative intent.
While the Evidence Act does not expressly contemplate AI-generated evidence, its provisions on electronic evidence could have provided a useful framework for determining the admissibility of such material. However, the Act appears to predicate the concepts of “document” and “information source” largely on the use of a computer or other tangible electronic device. AI systems, particularly those capable of generating autonomous outputs through complex machine-learning processes, do not fit neatly within this traditional categorisation. Consequently, the existing statutory framework may be inadequate to address the unique evidentiary challenges posed by AI-generated content, including questions relating to authorship, reliability, authenticity, and accountability.
Generative AI operates on principles that differ significantly from the assumptions embedded in Section 84, raising questions about how existing admissibility requirements should be applied to AI-generated content. The section appears to contemplate a human operator supplying information to a computer in the ordinary course of business activities. By contrast, modern AI systems generate outputs through complex machine learning, processes trained on vast datasets and models, often producing results that cannot easily be traced to a specific source of information.
The “black box” nature of many large language models and image generators may make it difficult to demonstrate that a system was “operating properly” within the contemplation of Section 84, particularly where the basis for a specific output cannot readily be explained. Similarly, questions arise as to whether AI-generated outputs can properly be regarded as information derived from data supplied in the ordinary course of activities, as envisaged by the Act.
The four conditions under Section 84(2) illustrate the challenge. First, AI systems may not always be used as part of a regular and established business process. Second, generative AI often relies on training data and model architecture that extend far beyond information supplied by the party tendering the evidence. Third, the tendency of AI systems to produce inaccurate or fabricated outputs commonly referred to as “hallucinations” raises difficult questions about what constitutes proper operation. Fourth, AI-generated outputs are frequently probabilistic compositions rather than direct reproductions or derivations of identifiable source material.
These observations do not necessarily mean that AI-generated evidence can never satisfy Section 84. Rather, they highlight the uncertainty involved in applying a provision designed for conventional electronic records to a technology that functions in fundamentally different ways.
Beyond these structural concerns lies a more immediate practical challenge. Under the current framework, electronic evidence that is not effectively challenged may be admitted, creating potential difficulties where parties lack the technical expertise necessary to identify AI-generated manipulation or fabrication. As AI-generated content becomes increasingly sophisticated, the risk of highly convincing but inaccurate evidence entering judicial proceedings becomes more pronounced.
The stakes are particularly high in electoral litigation, where digital evidence often plays a decisive role. The Electoral Act contains no provisions specifically addressing AI-generated evidence. In disputes involving election results, campaign finance, voter inducement, or electoral misconduct, video recordings, digital records, and electronic communications are routinely tendered before the courts.
Consider a scenario in which a party presents a deepfake video purporting to show electoral officials manipulating ballot materials or AI-generated financial records intended to support allegations of unlawful inducement. Under the current legal framework, the opposing party would bear the burden of challenging the authenticity of such evidence. Yet Nigerian law presently provides no specialised statutory framework for assessing AI-generated content, and courts may increasingly need to rely on expert evidence and evolving forensic techniques when such issues arise.
In an environment where access to digital forensic expertise remains limited, the potential consequences for electoral justice are significant.
Comparative Perspectives
Nigeria is not alone in confronting these challenges, but several jurisdictions have moved more quickly to develop legal and regulatory responses.
The European Union’s AI Act, which entered into force in 2024[3], imposes transparency obligations in relation to AI-generated and manipulated content. Article 50 requires certain AI-generated content to be clearly disclosed and, in appropriate cases, made machine-detectable. The objective is to improve traceability and reduce the risk of deception[4].
In the United States, the legislative landscape has evolved rapidly. Since 2022, 46 states have enacted deepfake legislation, and the federal TAKE IT DOWN Act became law in May 2025[5]. As of mid-2025, 47 states have enacted laws addressing deepfakes, and the federal TAKE IT DOWN Act was signed into law on May 19, 2025[6]. On 2 May 2025, the US Judicial Conference Advisory Committee on Evidence Rules voted 8–1 in favour of proposed Rule 707, Machine-Generated Evidence, which would subject AI and machine-generated evidence offered without an accompanying expert witness to the same reliability standards as expert testimony under Rule 702. The Committee on Rules of Practice and Procedure approved the proposed rule for public comment on 10 June 2025, and the public comment period closed on 16 February 2026, after which the rule faces further review and potential revision before any final adoption. The proposal, nonetheless, signals a clear direction, that AI-generated evidence will not be admitted in US federal courts without satisfying a structured reliability inquiry analogous to the Daubert standard for expert witnesses[7].
Although jurisdictions have adopted different approaches, a common trend is emerging, AI-generated content presents evidentiary concerns that differ from those associated with traditional electronic records. The most advanced responses combine legislative clarity, judicial training, and investment in digital forensic capabilities.
Nigeria’s response should proceed on three fronts.
First, legislative reform is necessary. Section 84 of the Evidence Act should be reviewed to determine whether a distinct framework is required for AI-generated evidence. Such a framework should focus not only on admissibility but also on authenticity, provenance, transparency, and reliability. Consideration should also be given to amendments to the Cybercrimes (Prohibition, Prevention, Etc.) Act to address AI-enabled fraud, deepfakes, and content provenance obligations.
Second, institutional capacity must be strengthened. The National Judicial Council, the Nigerian Bar Association, law enforcement agencies, and other relevant stakeholders should invest in training programs that equip judges, lawyers, and investigators to understand the evidentiary implications of artificial intelligence. As AI becomes more prevalent, technological literacy will become an increasingly important component of effective legal practice.
Third, Nigeria should consider developing a broader AI governance framework that incorporates transparency and traceability requirements for AI-generated content, particularly where such content may be used in legal proceedings. A risk-based regulatory approach, informed by international best practices while adapted to local realities, could help reduce the likelihood of AI-generated misinformation entering the justice system.
Conclusion
Section 84 of the Evidence Act was a landmark provision when enacted in 2011, and the 2023 amendments represented important progress in modernising Nigeria’s approach to electronic evidence. Yet technological development has not stood still. Generative AI has introduced new questions about authenticity, reliability, and evidentiary integrity that the current framework does not directly address.
The question of whether Nigeria is fully prepared for the challenges posed by AI-generated evidence remains open. What is clear, however, is that existing legal frameworks were developed before the emergence of generative AI and may require targeted reforms to preserve confidence in the administration of justice.
AI-generated evidence is no longer a purely future concern. As artificial intelligence becomes more accessible and more sophisticated, Nigerian courts, legislators, regulators, and legal practitioners will increasingly be called upon to address issues that existing laws only partially contemplate. The sooner those conversations translate into meaningful reform, the better positioned the legal system will be to safeguard the integrity of judicial proceedings in the age of artificial intelligence.
References
- (1969) 1 NMLR 194
- (2013) 4 NWLR (Part 1345) 534-594
- European Parliament (2024) EU AI Act: first regulation on artificial intelligence. Available at: https://www.europarl.europa.eu/topics/en/article/20230601STO93804/eu-ai-act-first-regulation-on-artificial-intelligence (Accessed: 11 June 2026)
- Blackbird.AI (2024) Deepfake detection required under the EU AI Act. Blackbird.AI Compass. Available at: https://blackbird.ai/blog/deepfake-detection-required-eu-ai-act-blackbird-ai-compass/ (Accessed: 11 June 2026).
- Jones Walker LLP (2024) Deepfakes as a service meets state laws governing synthetic media in a fragmented regulatory landscape. AI Law Blog. Available at: https://www.joneswalker.com/en/insights/blogs/ai-law-blog/deepfakes-as-a-service-meets-state-laws-governing-synthetic-media-in-a-fragmente.html?id=102m20g (Accessed: 11 June 2026).
- Shibolet & Co. (2025) EU AI Act: first draft code of practice on AI-generated content transparency. Available at: https://www.shibolet.com/en/eu-ai-act-first-draft-code-of-practice-on-ai-generated-content-transparency/ (Accessed: 11 June 2026).
- The National Law Review (2024) Proposed Rule 707 targets AI-crafted evidence. Available at: https://natlawreview.com/article/proposed-rule-707-targets-ai-crafted-evidence (Accessed: 11 June 2026).