Deepfakes and the Trial: Admissibility Challenges

Introduction

Recent months have seen a surge in reports that convincingly realistic ‘deepfake’ videos have been used to allege wrongdoing against public figures and private citizens alike. Courts in India and other Commonwealth jurisdictions are now confronting the possibility that audio-visual material admitted as evidence may have been synthetically generated or altered. This development matters because criminal trials hinge on reliable proof; the admission of manipulated digital media threatens the rights of accused persons, the integrity of prosecutions and the public’s confidence in adjudication.

Legal Background

The law governing electronic evidence in India centres on the Indian Evidence Act, particularly sections introduced to address electronic records (notably sections 65A and 65B). Since the Supreme Court’s decision in Anvar P.V. v. P.K. Basheer (2014), courts have required authentication of electronic records—commonly by a certificate under section 65B—before admitting them as evidence. Subsequent jurisprudence, including Arjun Panditrao Khotkar v. Kailash Kushanrao Gorantyal (2020), has elaborated the circumstances in which secondary evidence and certificates are necessary, but has not eliminated the need for rigorous proof of provenance. High Courts (e.g., Gurwinder Singh v. State of Punjab (Punjab-Haryana HC, 2024) and Gaurav Bhatia v. Naveen Kumar (Delhi HC, 2024)) have emphasised the role of expert digital forensics in establishing whether a recording is “genuine and untampered.”

Other doctrinal safeguards derive from criminal procedure and human rights principles: the accused’s right to a fair trial and to confront or test the prosecution’s evidence, and Article 21 protections over privacy and reputation when illicitly produced material is circulated. Where intermediaries or platforms host suspected deepfakes, rules on intermediary liability and takedown (in both domestic and comparative contexts) intersect with evidentiary questions.

Critical Analysis

Applying existing law to the recent wave of deepfake allegations exposes several pressure points. First, admissibility: courts must determine whether a contested clip is an original electronic record, a copy, or a fabricated artifact. Section 65B’s certificate (or equivalent expert proof) remains the primary mechanism to show that a digital file is what it purports to be. In practice, however, certificates alone may be insufficient where synthetic generation leaves no clear provenance trail. High courts have therefore required forensic proof beyond a bare certificate—hash values, metadata analysis, server logs, source-device extraction reports and chain-of-custody testimony by competent experts. The Punjab-Haryana High Court in Gurwinder Singh reiterated that an expert must “prove the digital version to be genuine and untampered.” That standard aligns with Delhi High Court guidance in Gaurav Bhatia, which scrutinised the interplay between the recording and corroborative evidence.

Second, the role of expert evidence: deepfakes are by design intended to fool human perception. Digital forensics experts must demonstrate the methodology used to detect manipulations (machine-learning artifacts, inconsistent lighting/shadow, frame-level interpolations, codec anomalies). Courts should be prepared to hear competing expert reports; trial judges must then decide weight and admissibility. The difficulty is that both sides may produce equally technical and persuasive experts, pushing the court into making complex technical credibility assessments. Adopting an evidentiary gatekeeping approach—admitting the material for limited purposes only, or ordering a neutral, court-appointed forensic analysis—may be necessary to preserve fairness.

Third, procedural safeguards and remedies: where the prosecution relies on a contested deepfake to secure arrest or detention, the risk of irreparable prejudice is acute. Courts should scrutinise warrants and remand decisions that rest primarily on unverified digital media. Interim remedies may include urgent disclosure orders to platforms for original-source logs, preservation orders preventing deletion, and ad hoc appointment of independent technical assessors. Comparative administrative jurisprudence—where courts have ordered expedited preservation from platforms—offers a pragmatic template.

Finally, public-interest and intermediary issues: platforms hosting deepfakes can be loci of evidence preservation but also sources of contamination (copies multiply, metadata lost). Judicial orders should balance platform obligations (to preserve logs and cooperate with lawful process) against due-process protections (avoiding overbroad disclosure demands and protecting privacy). The evolving body of interlocutory orders (e.g., recent Delhi High Court orders addressing platforms and synthetic content) demonstrates this tension.

Opinion & Outlook

Practical and doctrinal adaptations are required. First, courts should standardise a minimum technical protocol for handling suspected synthetic media: emergency preservation notices to platforms, mandatory neutral forensic analysis on contested recordings where admissibility is determinative, and clearer rules on the contents of Section 65B-style certificates (including explicit provenance and hash-chain documentation). Second, judicial training and specialist technical panels would improve gatekeeping—judges should be equipped to evaluate expert methodologies rather than only the expert’s credentials.

Legislatively, incremental reform is preferable to sweeping bans: amendments to the Evidence Act or procedural codes could define “authentication” in the context of synthetic media and set out preservation and disclosure duties for intermediaries. Internationally, judicial coordination and reliance on consensual technical standards (hashing, timestamping, provenance registries) may help reconstruct chains of custody where original devices are unavailable.

There is also a normative choice: whether to treat deepfakes as a new evidentiary species requiring bespoke rules, or to fold them into existing principles of provenance, authenticity and reliability. The safer path, and one consistent with rule-of-law values, is to adapt existing doctrines while adding practical procedural tools to address the unique technical challenges deepfakes present.

Conclusion

Deepfakes pose a real threat to the integrity of criminal fact-finding. Indian precedent already underscores the necessity of rigorous authentication and expert proof for electronic records; recent high-court decisions affirm that courts will not admit digital media without demonstrable proof of genuineness. To protect the right to a fair trial and preserve public confidence, courts, legislators and platforms must collaborate: standardise preservation and authentication practices, expand technical capacity within the judiciary, and ensure that evidentiary doctrines keep pace with synthetic-media technologies.