The year 2026 has become a defining moment for evidentiary law in India. Courts across the nation are grappling with a surge in cases involving deepfakes, automated surveillance reports, and AI-generated documents. The legal fraternity was shaken when the Supreme Court flagged the “alarming” trend of lawyers filing petitions citing non-existent judgments created by AI tools. Consequently, the question of AI-generated evidence admissibility BSA 2023 has moved from academic debate to urgent courtroom reality.
This shift coincides with the transition from the colonial-era Indian Evidence Act, 1872, to the Bharatiya Sakshya Adhiniyam (BSA) 2023. The new law expressly recognizes electronic and digital records within its definition of “document” under Section 2(1)(d). However, a core legal dilemma remains. Lawyers and judges must now determine “authenticity” and “authorship” when the creator is not a human, but an algorithm.
The Rise of Algorithmic Evidence
Indian courts are encountering AI in various forms in 2026. For instance, the Delhi High Court recently stayed a ₹1.08 crore decree. This decision came after noting a “prima facie impression of AI use” in a trial court judgment [^25]. Similarly, the Gujarat High Court addressed a Public Interest Litigation regarding the threat of AI-generated deepfakes to public order [^9]. These cases highlight the urgency for clear legal standards.
The Legal Framework: BSA 2023 and Electronic Records
The Bharatiya Sakshya Adhiniyam 2023 was designed to address technological advancements missing from the 151-year-old Evidence Act. Its primary tool for digital evidence is Section 63, the successor to Section 65B of the old Act. Understanding Section 63 is essential for any litigator dealing with AI-generated evidence admissibility BSA 2023.
Section 63 BSA 2023: The New Gold Standard
Section 63(1) states that any information in an electronic record, whether printed or stored, shall be deemed a document [^1]. However, this admissibility is subject to strict conditions. The statute requires that four cumulative conditions be satisfied regarding the computer’s regular use and operation [^3]. Furthermore, Section 63(4) introduces a rigorous certification requirement.
Unlike the old law, BSA 2023 mandates a specific two-part certificate format. Part A requires the party to disclose device details and, crucially, the hash value of the electronic record. Part B must be certified by an expert [^12]. The Supreme Court has explicitly upheld the constitutional validity of these requirements in 2026.
Hash Values as Electronic Fingerprints
In a landmark ruling, the Supreme Court in Pune Bar Association v. Union of India dismissed challenges to Section 63(4) [^6]. The Court observed that electronic records are susceptible to continuous alteration. Therefore, the mandatory disclosure of hash values serves a critical purpose.
The Bench, comprising CJI Surya Kant and Justice Joymalya Bagchi, stated that a hash value is synonymous with an “electronic fingerprint” [^18]. It provides a sure way of identifying and verifying digital data. This ruling fundamentally shapes the landscape of AI-generated evidence admissibility BSA 2023.
Supreme Court’s Landmark Interpretation on AI-Authored Content
The Supreme Court’s 2026 judgments have provided crucial clarity on how courts must treat AI outputs. The judiciary is increasingly wary of the “Black Box” problem, where AI generates outputs without transparent logic. This raises the question: Can code be cross-examined?
The “Algorithmic Hearsay” Dilemma
Legal scholars have coined the term “Algorithmic Hearsay” to describe this issue [^16]. When an AI tool like ChatGPT or a predictive policing algorithm produces an output, it acts as a “statement.” However, the traditional hearsay rule requires a human declarant who can be cross-examined. Since an algorithm cannot be sworn in, the Court must determine if the output is reliable evidence or inadmissible hearsay.
In Gummadi Usha Rani v. Sure Mallikarjuna Rao, the Supreme Court confronted this directly [^22]. The case revealed that a trial court had relied on four non-existent judgments generated by AI. The Court declared that basing decisions on fake, AI-generated precedents is not merely an error. Instead, it constitutes misconduct with legal consequences [^23]. Consequently, the Court urged the Bar Council of India to form an expert committee.
The “Black Box” Problem and Transparency
The Supreme Court’s reasoning in these cases emphasizes transparency. In Pune Bar Association, the Court specifically noted the threat posed by AI and deepfake tools [^6]. It observed that these developments intensify concerns about digital evidence reliability.
Moreover, the judiciary is signaling that AI outputs lacking verifiable training data or transparent algorithms may face high scrutiny. The “Black Box” problem—where not even developers fully understand the AI’s decision-making—creates an “analytical gap.” Courts may exclude such evidence if the gap between data and conclusion is too great [^16].
Authentication Standards: The “Deepfake Defense” and Burden of Proof
As AI tools become more sophisticated, proving the authenticity of evidence has become harder. The “Deepfake Defense”—claiming genuine evidence is AI-generated—is now a common litigation strategy. BSA 2023 and the IT Amendment Rules 2026 provide the framework to handle this.
IT Rules 2026 and Synthetic Information
The IT Amendment Rules 2026, effective from February 20, introduced the concept of “Synthetically Generated Information” (SGI) [^26]. SGI refers to audio-visual content created or altered by algorithms to appear authentic [^30]. Intermediaries are now obligated to deploy technical measures to prevent the dissemination of unlawful SGI. Furthermore, they must label permissible AI-generated content clearly. This regulatory shift impacts how courts view authentication.
Burden of Proof in AI Manipulation Cases
Under Section 106 BSA 2023, the burden of proof lies on the person with special knowledge of a fact [^11]. Therefore, if a defendant claims a video is a deepfake, they may bear the burden of proving that manipulation. In Vikas Vijay Nair v. State of Gujarat, the Gujarat High Court emphasized strict compliance with these new rules [^9]. The Court directed intermediaries to ensure that the integrity of digital evidence is preserved and verified.
Evidentiary Thresholds for Authentication
For lawyers, simply submitting a digital file is no longer sufficient. The Supreme Court’s interpretation mandates robust authentication. This includes verifying hash values (SHA256 is the preferred standard) and ensuring a complete chain of custody [^12]. The Court has clarified that the certificate requirement is not a mere formality. Rather, it is a substantive condition for admissibility.
Practical Implications for Litigators and Law Firms
The 2026 judicial landscape demands a new approach to digital evidence. Law firms must update their practices to ensure AI-generated evidence admissibility BSA 2023. Failure to do so risks adverse orders or even professional misconduct proceedings.
Checklist for Admitting AI Evidence
To successfully admit AI-generated or electronic evidence, lawyers must follow a strict protocol:
Generate Hash Value: Use tools like WinMD5Free to generate an SHA256 hash value immediately upon acquiring the digital evidence. Complete Part A Certificate: Record device details (Make, Model, Serial No., MAC ID) and the hash value. Obtain Part B Certification: Engage an Examiner of Electronic Evidence under Section 79A IT Act, or a recognized cyber forensic expert [^12]. Attach Hash Report: The hash report must accompany the certificate at every instance of submission [^3].
Strategies to Challenge AI Evidence
Challenging AI evidence requires technical and legal acumen. Lawyers can question the “Black Box” nature of the algorithm. Additionally, they can demand disclosure of the training data, known biases, and error rates [^16]. For example, invoking the “parity of arms” principle ensures the defense has access to technical experts.
Lawyers should also scrutinize the methodology of the AI tool. Specifically, if the analytical gap between input and output is unclear, the evidence may be excluded.
The Role of Expert Witnesses
Section 39 BSA 2023 recognizes opinions of experts specifically skilled in science or art [^15]. The Supreme Court has broadened this to include those with special expertise in computer science or cyber forensics [^18]. Expert witnesses now play a pivotal role in explaining AI mechanics to judges. They can validate the hash value integrity or debunk claims of deepfake manipulation. Therefore, building a network of trusted forensic experts is vital for modern litigation practice.
Future Outlook: Statutory Reforms on the Horizon
While the Supreme Court’s 2026 rulings have clarified the law, the legal framework is still evolving. The rapid pace of AI development may soon outstrip the current provisions of BSA 2023.
The Need for a “Digital Evidence Act”
Legal scholars argue that BSA 2023 treats AI primarily as an electronic record focused on storage [^14]. However, it lacks specific provisions for machine-generated expert opinion. There is a growing consensus for a dedicated “Digital Evidence Act” or amendments to BSA. Such reforms could mandate a “glass-box” approach, requiring algorithmic transparency for admissibility. This would align with global standards like the EU AI Act 2024, which classifies judicial AI systems as “high-risk” [^16].
Global Standards and India’s Path
Comparatively, the US uses the Daubert standard for scientific evidence, focusing on testability and error rates. China has integrated AI directly into its “206 systems” for judicial decision-making. India’s approach, balancing strict certification with judicial oversight, seems to follow a middle path. The Supreme Court has already indicated a push for “sovereign large language models” to mitigate hallucination risks [^22]. This suggests a future where government-verified AI tools may have a presumption of reliability.
Conclusion
The Supreme Court’s 2026 rulings have firmly established that AI-generated evidence is admissible, but only under strict conditions. The AI-generated evidence admissibility BSA 2023 standards prioritize authenticity, transparency, and expert verification. For Indian lawyers, the message is clear: neglecting hash values or certification protocols is no longer an option. As deepfakes and AI tools proliferate, the integrity of the judicial process depends on rigorous evidentiary compliance.
Stay ahead of evolving legal tech standards. Use LawSathi’s AI-powered research tools to instantly analyze landmark judgments and draft evidence submissions compliant with BSA 2023. Start your free trial today.
