Deepfakes And the Crisis of Authenticity: Challenges to the Admissibility and Reliability of Electronic Evidence in Modern Courts

CONTRIBUTOR: ETIENO MARTIN UKANGA

Introduction

The growing acceptance of electronic evidence in judicial proceedings has significantly transformed how courts establish facts and determine liability. Digital photographs, video recordings, and audio files are now routinely admitted as evidence and often carry considerable probative value, particularly where they appear to provide a direct account of disputed events. However, the emergence of deepfake technology, driven by advanced artificial intelligence (AI) techniques such as deep learning and generative adversarial networks (GANs), poses a serious threat to the trust traditionally placed in digital evidence.

Deepfakes enable the creation of highly realistic yet entirely fabricated images, videos, and audio recordings that portray individuals engaging in acts they never performed or making statements they never made. This technological capability presents a significant legal challenge: how should courts respond when manipulated electronic evidence is so convincing that it appears authentic and aligns with surrounding circumstances? Consider a situation in which two individuals are falsely depicted in a video attending a particular location and committing a wrongful act. If ancillary evidence, such as location data, witness testimony, or evidence of prior association, appears to corroborate the recording, disproving the allegation may become considerably more difficult.

In such circumstances, the ability of the affected individuals to deny involvement may be undermined, not because the evidence is genuine, but because it is persuasive. Courts, which traditionally rely on the apparent reliability and authenticity of electronic records, may find it increasingly difficult to distinguish between genuine and manipulated content.

This development raises a profound legal concern: the possibility that courts may rely on convincing but fabricated electronic evidence, thereby compromising the integrity of the judicial process. This article examines the nature of deepfake technology, the principles governing the admissibility, weight, and reliability of electronic evidence, and the challenges that deepfakes pose to evidentiary authenticity. It further highlights the urgent need for stricter verification standards and procedural safeguards to preserve the integrity of judicial proceedings in the digital age.

The Definition and Concept of Deepfake

A deepfake is a form of synthetic media, typically a video, audio recording, or image, that has been manipulated or generated using AI to create the false impression that a person said or did something they never actually said or did. The term deepfake combines “deep learning,” a subset of AI, and “fake.” Through this technology, users can generate highly convincing audio and visual content depicting real individuals engaging in fabricated speech or conduct.

Advancements in machine learning have significantly increased the sophistication of deepfake technology, making manipulated content more realistic and increasingly difficult to detect. The accessibility of AI-powered tools has also enabled both sophisticated and unsophisticated actors to rapidly and on a large scale create and disseminate deepfakes. As a result, deepfakes present serious challenges to information integrity, privacy, security, and the administration of justice.^[1]

Developers typically create deepfakes using a machine learning technique known as a GAN.^[2] A GAN consists of two competing AI models that continuously improve through interaction:

• The generator creates fabricated content
• The discriminator evaluates the content and attempts to determine whether it is genuine or artificial.

Through repeated iterations, both models improve their performance until the generated content becomes highly realistic and indistinguishable from authentic material. This process explains why modern deepfakes often appear remarkably convincing.

Deepfakes commonly take the following forms:

1. Face-Swapped Videos

This technique replaces a person’s face in a video with another individual’s, creating the false appearance that the latter participated in the recorded event.

2. AI-generated Voice Recordings (Voice Cloning)

AI systems replicate a person’s voice patterns, tone, and speech characteristics to produce fabricated audio recordings that closely resemble genuine speech.

3. Fully Synthetic Persons or Scenes

AI can generate images or videos of individuals, environments, or events that do not exist in reality. These creations may appear entirely authentic despite being wholly artificial.

Criminals and fraudsters have increasingly exploited deepfake technology to perpetrate fraud, impersonation, extortion, misinformation campaigns, and other forms of criminal activity across the world.

Deepfakes and the Admissibility of Electronic Evidence Under Nigerian Law

The admissibility of electronically generated evidence, including audio recordings, video recordings, digital photographs, and potentially deepfakes, is principally governed by the Evidence Act 2011. While the Act recognises the growing importance of computer-generated evidence in modern litigation, it imposes specific conditions to ensure its authenticity, reliability, and integrity before a court may admit it.

Admissibility of Computer-Generated Evidence

Section 84 of the Evidence Act 2011 is the principal statutory provision governing the admissibility of electronic and computer-generated evidence in Nigeria. The provision recognises the increasing role of digital technology in modern communication and record-keeping while establishing safeguards to ensure the reliability of electronically generated evidence.

Key Requirements Under Section 84

For electronically generated evidence, such as videos, audio recordings, emails, and other digital records, to be admissible in evidence, the party seeking to rely on the evidence must satisfy the conditions stipulated in Section 84 of the Evidence Act 2011. These conditions include the following:

• The computer must have been used regularly to store or process information during the relevant period.
• The information contained in the electronic record must have been supplied to the computer in the ordinary course of business or other regular activities.
• The computer must have been operating properly throughout the material period, or any malfunction must not have affected the accuracy of the record.
• A certificate of compliance must accompany the electronic evidence. The certificate must identify:

1. the document sought to be tendered;
2. the manner in which it was produced; and
3. particulars of the computer or device used in generating the document.

Failure to comply with the requirements of Section 84 generally renders the electronic evidence inadmissible.

Admissibility of Video and Electronic Evidence

Nigerian courts recognise video recordings and other electronic materials as documentary evidence, provided they satisfy the applicable legal requirements. To be admissible, such evidence must:

• be relevant to the facts in issue in accordance with Sections 1–7 of the Evidence Act 2011;^[3]
• comply with the certification requirements under Section 84; and
• demonstrate authenticity and integrity.

The courts have gradually moved away from the traditional scepticism towards electronic evidence and now readily admit properly certified digital evidence. Nevertheless, admissibility does not automatically confer credibility. A distinction exists between the admissibility of evidence and the weight a court attaches to it. While admissibility concerns whether the court may admit the evidence, weight concerns its probative value and reliability.

Deepfake technology presents a significant challenge in this regard. A highly sophisticated deepfake video may satisfy the procedural requirements of Section 84 and therefore be admitted in evidence. However, despite its admissibility, the content may be entirely fabricated. This reality raises serious concerns about the reliability and evidentiary weight of electronic records and underscores the need for courts to adopt more rigorous authentication methods when evaluating digital evidence.

Danger of Deepfakes: Circumstantial Evidence, Deepfake Technology, and the Evidentiary Burden in Criminal Trials

In criminal proceedings, it is a well-established principle that a court may convict an accused solely on circumstantial evidence, provided that the evidence is credible and cogent and forms a complete chain that points irresistibly to the accused’s guilt. In Lori & Anor v State,^[4] the Supreme Court of Nigeria held:

“It is conceded that circumstantial evidence is very often the best evidence. It is said to be evidence of surrounding circumstances which by undesigned coincidence is capable of proving a proposition with the accuracy of mathematics.”

The Court further emphasised that describing evidence as circumstantial does not diminish its probative value. Likewise, in Abieke & Anor v State,^[5] the Court affirmed that circumstantial evidence must exclude every reasonable hypothesis other than the guilt of the accused. Similarly, in R v Teper, the Privy Council cautioned that courts must carefully scrutinise circumstantial evidence, as it may be fabricated to cast suspicion on an innocent person.

This principle, though sound, becomes increasingly problematic in the age of artificial intelligence, particularly with the emergence of deepfake technology. Consider a situation in which an individual is present at or near the scene of a crime but plays no role in its commission. If a deepfake video subsequently emerges portraying that individual as the perpetrator, and independent circumstantial evidence places the individual within the vicinity of the crime, the cumulative effect of the evidence may be highly prejudicial. In such circumstances, the evidentiary narrative may appear so compelling that it becomes exceedingly difficult for the accused to rebut the inference of guilt.

Under sections 84^[6] and 34 of the Evidence Act 2011 (Nigeria),^[7] electronically generated evidence, including video recordings, is admissible once the statutory requirements relating to authenticity and the proper functioning of the device used to produce the record have been satisfied. Nigerian courts have consistently maintained that compliance with section 84 is a condition precedent to admissibility. In Dickson v Sylva & Ors, the Court reaffirmed that a party seeking to tender electronically generated evidence must strictly comply with the requirements of section 84. Once admitted, the court proceeds to determine the weight to be attached to the evidence by considering its relevance, credibility, and the surrounding circumstances.

However, a significant doctrinal gap exists between admissibility and authenticity. A deepfake video may satisfy the formal requirements for admissibility while remaining fundamentally unreliable. The current legal framework primarily focuses on procedural requirements for admitting electronic evidence and does not adequately address the unique risks posed by AI-generated manipulations. Consequently, a court may admit a deepfake recording and accord it substantial probative value, particularly where it appears consistent with other circumstantial evidence.

Practical realities within Nigeria further compound this challenge. Effectively contesting the authenticity of sophisticated digital evidence often requires specialised digital forensic expertise. However, Nigeria currently has a limited pool of highly trained artificial intelligence specialists and digital forensic analysts. Many of these experts operate within private organisations or specialised government agencies such as the Economic and Financial Crimes Commission (EFCC) and the forensic units of the Nigeria Police Force. For many litigants, particularly defendants in criminal proceedings, access to such expertise remains constrained by cost, availability, and institutional limitations.

Consequently, an accused person who denies being the individual depicted in a deepfake video may face considerable difficulty obtaining expert evidence capable of exposing the manipulation. This reality raises concerns regarding the constitutional guarantee of fair hearing under section 36 of the Constitution of the Federal Republic of Nigeria 1999 (as amended), as disparities in access to forensic resources may create an uneven playing field. Furthermore, the judiciary may not yet possess sufficient technological or institutional capacity to independently evaluate the authenticity of highly sophisticated AI-generated evidence without relying extensively on expert testimony. As a result, courts may inadvertently place greater reliance on the apparent credibility of digital evidence than on its actual integrity.

An additional layer of complexity arises from what scholars have termed the “liar’s dividend,” a concept popularised by Bobby Chesney and Danielle Citron. The term describes a situation in which the existence of deepfake technology enables individuals to dismiss genuine evidence as fabricated, thereby undermining confidence in authentic digital records. Ironically, as courts and the public become increasingly aware of the dangers posed by deepfakes, individuals seeking to evade responsibility for their genuine words or actions may find it easier to deny their authenticity.^[8] The legal system therefore faces a dual threat: the risk of wrongful convictions based on fabricated evidence and the risk of wrongful acquittals resulting from unfounded claims that authentic evidence has been manipulated.

Accordingly, although Nigerian law permits convictions based on circumstantial and electronic evidence, the rapid advancement of artificial intelligence technology demands a corresponding evolution in legal standards, forensic capabilities, and judicial awareness. Without deliberate reforms, including enhanced authentication requirements, greater institutional investment in digital forensic expertise, and specialised judicial training on emerging technologies, the criminal justice system may become increasingly vulnerable to miscarriages of justice arising from sophisticated forms of digital deception.

Proposed Recommendation, Reforms and the Way Forward in Addressing Deepfake Evidence

The challenges posed by deepfake technology require urgent and deliberate reforms within the Nigerian criminal justice system. Although the Evidence Act 2011 provides a framework for the admissibility of electronic evidence, its provisions do not adequately address the unique challenges presented by artificial intelligence-generated content. As deepfake technology becomes increasingly sophisticated, the legal system must develop safeguards to prevent courts from relying on manipulated digital evidence. Accordingly, the following reforms merit consideration.

1. Stronger Digital Forensic Capacity

Digital forensics involves the application of scientific methods to identify, preserve, analyse, investigate, and detect alterations to electronic data and to present such evidence in a legally admissible form.^[9] The increasing prevalence of deepfake technology underscores the need for greater investment in digital forensic capabilities across the criminal justice system.

Courts, law enforcement agencies, and prosecutorial authorities should be equipped with modern forensic laboratories, advanced AI-detection tools, and adequately trained digital forensic specialists. Judicial officers, prosecutors, defence counsel, and investigators should also receive continuous training on emerging technologies and their evidential implications. Such investment is necessary because the authenticity of electronic evidence, particularly audio and video recordings, cannot be presumed in an era of sophisticated digital manipulation. Robust forensic capacity will assist courts in evaluating the integrity and reliability of digital evidence and reduce the risk of wrongful convictions or acquittals arising from fabricated content.^[10]

2. Stricter Authentication Standards by Expert Evidence

Given the increasing sophistication of deepfake technology, courts should adopt more rigorous authentication requirements before admitting electronic evidence that is susceptible to manipulation. While compliance with section 84 of the Evidence Act 2011 remains necessary, it may no longer be sufficient in cases where the authenticity of digital content is in dispute.

A strengthened authentication framework should require independent expert verification of electronic evidence where credible allegations of manipulation arise. Such verification may include:

Technical Verification

This involves the use of digital forensic tools and expert analysis to detect signs of manipulation, including inconsistencies in pixels, lighting, shadows, facial movements, audio patterns, compression artefacts, and metadata. Technical examination may also reveal imperceptible anomalies indicative of artificial generation or alteration.

The increasing importance of forensic verification reflects the reality that visual authenticity does not necessarily equate to evidential authenticity. Consequently, courts should place greater reliance on scientifically verifiable authentication methods when assessing the reliability of digital evidence in the age of artificial intelligence.

Metadata and Chain of Custody

Courts should require parties relying on digital evidence to establish a clear chain of custody and provide metadata capable of demonstrating when, where, and how the electronic record was created, transmitted, stored, and preserved. Proper documentation should also show that the file has not been altered from the time of its creation to its presentation in court. Where allegations of manipulation arise, or where the circumstances reasonably suggest the possibility of alteration, courts should require verification by qualified digital forensic experts or artificial intelligence specialists. This approach accords with emerging international best practices for the authentication of complex digital evidence.^[11]

Corroboration Requirement

Electronic evidence susceptible to deepfake manipulation should not stand alone as proof of material facts. Courts should require corroborative evidence from independent sources, such as eyewitness testimony, geolocation data, communication records, metadata analysis, or other forensic findings. Requiring corroboration would reduce the risk that courts rely on fabricated digital content and strengthen the overall reliability of the evidentiary process.

Reconsidering the Burden of Proving Authenticity

The Evidence Act 2011 may require legislative reform to address the unique challenges posed by deepfakes. Where a party raises a credible challenge to the authenticity of a digital recording on the ground that it may be a deepfake, the law could require the party relying on the evidence to establish its authenticity through additional verification measures.^[12]

However, any reform in this regard must be carefully framed to avoid undermining the constitutional presumption of innocence or improperly shifting the legal burden of proof in criminal proceedings. Rather than imposing a true reverse burden, the law may require the proponent of the evidence to satisfy enhanced authentication requirements where a genuine question of authenticity has been raised.

Statutory Requirement for Expert Verification

The legislature should consider introducing specific provisions requiring forensic verification of disputed digital evidence, particularly audio and video recordings whose authenticity is challenged. Such provisions may require certification or verification by qualified digital forensic experts before courts attach substantial weight to the evidence. This would reduce judicial speculation, promote consistency in decision-making, and enhance confidence in the reliability of electronic evidence.

Regulatory Measures

Addressing the risks associated with deepfakes requires more than evidential reform. Technology companies that develop or deploy artificial intelligence tools should implement safeguards to promote transparency and accountability in the creation of AI-generated content.^[13] Such measures may include disclosure requirements, digital watermarking, content provenance systems, and mechanisms for identifying artificially generated media.

In addition, the government should consider introducing specific criminal offences targeting the malicious use of deepfake technology. Such offences could criminalise the creation or dissemination of deepfake content for the purpose of harming, intimidating, threatening, impersonating, or defrauding another person. Several jurisdictions have already adopted legislative measures addressing these concerns, and Nigeria may draw useful lessons from such developments when formulating an appropriate regulatory framework.^[14]

Judicial Caution

Courts must exercise increased caution when assessing the authenticity and reliability of electronic evidence in the era of deepfake technology. Useful guidance may be drawn from international developments such as the European Union’s Electronic Identification, Authentication and Trust Services (eIDAS) Regulation, which establishes a framework for the authentication and verification of electronic transactions and recognises various forms of digital evidence,^[15] including electronic signatures, electronic seals, and electronic time stamps. Under the Regulation, electronic records are not to be denied legal effect or admissibility solely because they are in electronic form.^[16]

The eIDAS framework also underscores the importance of establishing institutional mechanisms to assist courts in resolving questions of the authenticity of digital evidence. Through the eIDAS Cooperation Network, guidance may be provided on issues relating to authentication, verification, and the assessment of disputed electronic materials. Although the Nigerian legal system differs from that of the European Union, the underlying principles of reliability, verification, and technological competence remain instructive.^[17]

Accordingly, Nigeria should consider developing institutional guidelines and judicial protocols for evaluating electronic evidence, particularly where allegations of deepfake manipulation arise. Such measures would assist courts in assessing the authenticity of digital materials and promote greater consistency in judicial decision-making.

Conclusion

Deepfake technology poses a significant challenge to the admissibility, authenticity, and reliability of electronic evidence in Nigeria. Although the Evidence Act 2011 recognises and regulates the admissibility of electronic evidence, its provisions were enacted before the emergence of sophisticated AI-generated manipulations capable of producing highly realistic fabricated content. As a result, compliance with the formal requirements of admissibility may not always guarantee the authenticity or reliability of digital evidence.

To address this challenge, courts must adopt a more rigorous approach to evaluating electronic evidence, particularly when manipulation is at issue. Strengthening digital forensic capacity, enhancing authentication standards, promoting expert verification, and improving judicial awareness of emerging technologies will be critical to preserving the integrity of the criminal justice system. Ultimately, the law must evolve alongside technological developments to ensure that advances in artificial intelligence do not undermine the administration of justice.

Reference

1. Chesney, B. & Citron, D., “Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security” (2019) 107 California Law Review 1753, at pp. 1785–1786; https://static1.squarespace.com/static/640d6616cc8bbb354ff6ba65/t/644f7199fe88295552b00f0e/1682928026258/2+-+Chesney+%2B+Citron.pdf – Accessed 19^th May, 2026. ↑
2. https://lab.slv.vic.gov.au/glossary/generative-adversarial-network ↑
3. Evidence Act 2011, S. 1-7. ↑
4. (1980) LPELR-1794 (SC) 8–11 ↑
5. (1975) LPELR- 8042 (SC)9–11 ↑
6. Ibid. S.84 ↑
7. Evidence Act 2011 (S.34) ↑
8. Ibid, n1. ↑
9. James R. Lyle, Barbara Guttman, John M. Butler, Kelly Sauerwein, Christina Reed, Corrine E. Lloyd (2022) Digital Investigation Techniques: A NIST Scientific Foundation Review. (National Institute of Standards and Technology, Gaithersburg, MD), NIST Internal Report (IR) NIST IR 8354. https://doi.org/10.6028/NIST.IR.8354 ; https://nvlpubs.nist.gov/nistpubs/ir/2022/NIST.IR.8354.pdf -Accessed 20th May, 2026. ↑
10. NIST-DIGITAL EVIDENCE AND DIGITAL FORENSIC https://www.nist.gov/digital-evidence – Accessed 20th May, 2026. ↑
11. Bateman, ‘Deepfakes and Synthetic Media in the Financial System: Assessing Threat Scenario’; Maria Pawelec, Cora Bieß, and Alexander Orlowski, ‘Ethisch Und Sozial Wünschenswerte Technikgovernance Fördern,’ Eberhard Karls Universität Tübingen, 2021- Accessed 20^th May, 2026. ↑
12. European Parliament: Directorate-General for Parliamentary Research Services, Tackling deepfakes in European policy, European Parliament, 2021, https://data.europa.eu/doi/10.2861/325063– file:///C:/Users/User/Downloads/tackling%20deepfakes%20in%20european%20policy-QA0221930ENN.pdf Accessed 20th May, 2026. ↑
13. Article 52 (3) European Commission, ‘Proposal for a Regulation Laying down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act).- Accessed 20^th May, 2026. ↑
14. Ibid, n13. ↑
15. Regulation (EU) No 910/2014 of the European Parliament and of the Council of 23 July 2014 on electronic identification and trust services for electronic transactions in the internal market and repealing Directive 1999/93/EC. See also: Proposal for a Regulation of the European Parliament and of the Council on European Production and Preservation Orders for electronic evidence in criminal matters, COM/2018/225 final – 2018/0108 (COD) – Accessed 20^th May, 2026. ↑
16. See: articles 25, 35, 41, 43 and 46 of the EU eIDAS regulation. ↑
17. European Commission, ‘Communication on Tackling Online Disinformation: A European Approach.’- Accessed 20^th May, 2026. ↑