Deepfake Evidence in Indian Courts: Legal Challenges, Admissibility, Authentication & Burden of Proof

The Courtroom Deception That Changed Everything

Picture this: A Delhi courtroom on an ordinary Tuesday morning. Weeks earlier, the prosecution had filed its evidence: a crystal-clear video, duly accompanied by a Section 63 certificate and hash value, admitted on record before the trial commenced.

Table of Contents

Now, in open court, the video is played for the witness under examination. In it, the accused—a corporate executive—is shown accepting a substantial bribe from a known crime syndicate member. The voice matches perfectly. His distinctive mannerisms, his facial expressions of greed, the way he checks his surroundings nervously—all captured in unflinching detail.

The defence counsel rises during cross-examination and makes an unprecedented claim:

“Your Honour, this video is a deepfake. My client never took any bribe. This is artificial intelligence fabrication.”

The courtroom falls silent.

What follows is not a straightforward verdict on guilt or innocence, but rather a meta-trial about the very nature of evidence itself.

Critical Questions Before the Court

Who carries the burden of proof?
How can either side authenticate what appears authentic?
What forensic tools can courts rely upon when technology itself has become both accuser and defendant?

This scenario, while hypothetical, encapsulates a crisis that Indian courts are beginning to confront with alarming urgency.

Why This Scenario Matters

Welcome to the era where seeing is no longer believing and where the age-old maxim verba scripta manent (the written word endures) has become dangerously unreliable.

This article examines the legal challenges that deepfake evidence poses to the Indian judicial system, the shortcomings of our existing legal frameworks, and the international measures that India may draw upon to protect the integrity of courtroom proceedings.

Key Issues Discussed in This Article

Topic	Focus
Deepfake Evidence	Challenges posed by AI-generated audio and video in court proceedings.
Burden of Proof	Determining who must establish authenticity or fabrication.
Digital Authentication	The role of Section 63 certificates, hash values, and forensic verification.
Judicial Challenges	How Indian courts can evaluate evidence in the age of artificial intelligence.
Comparative Legal Approaches	International measures that India may adopt to safeguard judicial integrity.

Understanding Deepfakes: Technology Meets Deception

The term “deepfake” is a portmanteau of “deep learning” and “fake”, a technology that sits at the intersection of artificial intelligence and sophisticated media manipulation. Deepfakes are artificially generated or manipulated videos, audio, or images created using generative adversarial networks (GANs) and deep learning algorithms, designed to appear so authentic that distinguishing them from genuine content becomes nearly impossible, even to the trained eye.¹

How Deepfake Technology Works

What makes deepfakes particularly insidious is their sophistication. Unlike crude video forgeries of yesteryear, modern deepfake technology learns from thousands of hours of footage, photographs, and audio recordings. Feed an AI system enough images of a person’s face and sufficient recordings of their voice, and it can generate content where that person appears to say or do things they never actually said or did. The technology replicates not just facial features but also nuances like the slight asymmetry in a smile, the particular cadence of speech, and the micro-expressions that ordinarily signal authenticity.²

Why Deepfakes Are Dangerous

Artificial intelligence can generate highly realistic videos, images, and audio.
Deepfakes imitate facial expressions, voice patterns, and body language.
Modern AI makes fake content extremely difficult to distinguish from genuine recordings.
The technology creates significant legal, social, and evidentiary challenges.

The Rashmika Mandanna Deepfake Case in India

In November 2023, India received a rude awakening to this technology’s potential for misuse. A deepfake video of Bollywood actress Rashmika Mandanna went viral on social media platforms, appearing to show her in a sexually explicit dance sequence. The video was entirely fabricated using AI-generated imaging techniques. The creator, a 23-year-old engineer from Andhra Pradesh, was apprehended within weeks and charged under multiple sections of the Information Technology Act and the Indian Penal Code. Yet his arrest marked merely the beginning of understanding how deeply this technology could penetrate Indian legal and social systems.³

The Indian Legal Framework: Existing Provisions and Stark Gaps

India’s approach to regulating deepfakes is, regrettably, scattered across multiple legislative enactments, none of which were drafted with this technology in mind. The primary frameworks are the Indian Evidence Act (now replaced by the Bharatiya Sakshya Adhiniyam, 2023), the Information Technology Act, 2000, and various provisions of the Indian Penal Code (now, the Bharatiya Nyaya Sanhita 2023).

Primary Laws Governing Deepfakes in India

Legal Framework	Relevance to Deepfakes
Bharatiya Sakshya Adhiniyam, 2023	Governs admissibility and authentication of electronic evidence.
Information Technology Act, 2000	Provides offences relating to electronic records and cyber activities.
Bharatiya Nyaya Sanhita, 2023	Contains criminal provisions applicable to fraudulent or harmful use of deepfakes.

The Bharatiya Sakshya Adhiniyam, 2023, and the Authentication Paradox

The Bharatiya Sakshya Adhiniyam (BSA), 2023, which came into force on July 1, 2024, replaced the Indian Evidence Act, 1872. Under Section 62 and Section 63 of the BSA, electronic records including digital videos and audio files are admissible as evidence, provided they meet certain authentication requirements. Section 63(4)(c) mandates that electronic evidence must be accompanied by a certificate detailing the source, the manner of production, and the particulars of the device involved in its creation.⁴

Electronic Evidence Certificate Requirements

The certificate now requires two components: Part A, completed by the party submitting the evidence, and Part B, completed by an expert. Section 63(4) of the BSA provides that the certificate must specify “the device’s identifying number, a description of the electronic record, and details of the manner in which it was produced”. Additionally, the BSA now mandates a hash value, a unique numeric identifier, to verify that the electronic record has not been altered. This represents a significant step forward from the older Section 65B of the Indian Evidence Act.⁵ However, and this is crucial, the framework presupposes that proper certification and hash verification can determine authenticity. For deepfakes, this assumption crumbles.

Key Authentication Requirements Under the BSA

Electronic records must be accompanied by a statutory certificate.
Part A must be completed by the party producing the evidence.
Part B must be completed by an expert.
The certificate must identify the device used.
It must describe the electronic record.
It must explain the manner in which the record was produced.
A hash value must be provided to verify that the electronic record has not been altered.

Procedural Authenticity vs. Factual Authenticity

Consider the problem: if a deepfake is created on a personal computer using legitimate software, it will produce a valid hash. The device that created it was indeed the device specified in the certificate. The chain of custody appears unbroken. From a purely procedural standpoint under Section 63 of the BSA, the evidence might appear admissible. Yet from a substantive standpoint, it is entirely fabricated. The law conflates procedural authenticity with factual authenticity, a dangerous equation when synthetic media is involved.⁶

Key Takeaways

The BSA strengthens procedures for admitting electronic evidence.
Hash values verify whether a file has been altered after creation.
A valid certificate does not establish that the content itself is genuine.
Deepfakes expose the distinction between procedural authenticity and factual authenticity.
This gap presents one of the greatest evidentiary challenges posed by AI-generated media.

The Information Technology Act, 2000: Fragmented Protections

The Information Technology Act contains several provisions that tangentially address deepfake creation and distribution, though none are specifically tailored to this technology.

Relevant Provisions Under the Information Technology Act

Provision	Scope	Relevance to Deepfakes
Section 66C	Identity theft	Criminalises identity theft, carrying imprisonment up to three years or a fine up to one lakh rupees.
Section 66D	Cheating by personation using computer resources	Addresses “punishment for cheating by impersonation using computer resources”, a provision that has been invoked in deepfake prosecutions, though it was originally designed for simpler forms of online fraud. 7
Section 66E	Privacy violations	Deals with privacy violations by “capturing, publishing or transmitting the image of a private area of a person without consent” and has been stretched to cover non-consensual intimate deepfakes, though the provision’s language around “capturing” creates ambiguity when the image was never actually captured but entirely generated by algorithms. 8
Sections 67 and 67A	Obscene and sexually explicit content	Apply to sexually explicit deepfakes, but require that the content be transmitted or published, not merely created.
Section 79	Safe harbor for intermediaries	Provides conditional safe harbour to intermediaries while requiring timely removal of harmful content.

Limitations of Existing IT Act Provisions

None of the provisions are specifically drafted to regulate deepfake technology.
Section 66D was originally intended for traditional online fraud rather than AI-generated impersonation.
Section 66E creates ambiguity because deepfake images may be entirely AI-generated rather than “captured”.
Sections 67 and 67A require publication or transmission, not merely creation.
Section 79 places significant compliance obligations on intermediaries.

Section 79 provides conditional safe harbour to intermediaries but places a heavy burden on them to identify and remove harmful content within specified timeframes, often 24 hours for sexual content. This is technologically challenging given the speed at which deepfake detection tools evolve and the difficulty in distinguishing deepfakes from authentic content even with specialised software. 9

The Bharatiya Nyaya Sanhita, 2023: Older Solutions for Modern Crimes

The newly enacted Bharatiya Nyaya Sanhita (Criminal Code) contains provisions on forgery (Section 248-256, previously IPC Section 465-475), defamation (Section 356, previously IPC Section 499), and cheating (Section 318-320, previously IPC Sections 415-420).

Criminal Provisions Used in Deepfake Prosecutions

Provision	Traditional Purpose	Challenge in Deepfake Cases
Forgery (Sections 248-256)	Altering authentic documents	Deepfakes involve creating entirely synthetic content rather than altering existing material.
Defamation (Section 356)	Protecting reputation	Requires first proving that the content is synthetic before reputational harm can be established.
Cheating (Sections 318-320)	Fraudulent deception	Requires proof of fraudulent intent, which may not exist where deepfakes are created for entertainment or political commentary.

Limitations of the Bharatiya Nyaya Sanhita

Forgery provisions focus on altered authentic documents rather than AI-generated synthetic content.
Defamation prosecutions require establishing both falsity and reputational harm.
Cheating provisions depend upon proving fraudulent intent.
Existing criminal provisions were not drafted with generative AI technologies in mind.

Deepfake creators have been prosecuted under these provisions, yet each has limitations. Forgery traditionally involves altering authentic documents; deepfakes involve creating something entirely synthetic. Defamation requires proving false statements caused reputational harm; a deepfake prosecution requires proving the synthetic nature of the content itself before defamation can even be addressed. Cheating requires proving fraudulent intent to deceive, an intent that may not always be present, particularly when deepfakes are created for entertainment or political commentary. 10

The Admissibility Crisis: Deepfakes as Evidence

The evidentiary challenges posed by deepfakes strike at the very heart of the judicial process. When deepfake evidence is introduced in court, several critical questions arise, none of which are adequately addressed by existing jurisprudence or statutory provisions.

Key Evidentiary Challenges Posed by Deepfakes

Difficulty in establishing authenticity.
Limitations of existing electronic evidence rules.
Problems relating to chain of custody.
Traceability of AI-generated content.
Burden of proving fabrication in court.

The Authenticity Conundrum

In the landmark case Anvar P.V. v. P.K. Basheer (2014), the Supreme Court established that electronic evidence must satisfy three criteria: it must be authentic, created in the ordinary course of business, and duly certified. The court held that a certificate under Section 65B (now Section 63 of the BSA) is mandatory for the admissibility of electronic evidence when the original is not produced.¹¹ However, this framework assumes that authenticity can be verified through procedural means checking the source, the storage medium, and the integrity of transmission.

Deepfakes fundamentally challenge this assumption. A deepfake certificate may be perfectly valid; the hash may match; the device specifications may be correct, yet the content itself is fabricated. The law essentially asks, “Did this file come from where you claim?” not “Is this file truthful?” For ordinary digital evidence, these questions are congruent. For deepfakes, they diverge catastrophically.¹²

Authenticity vs. Truthfulness

Traditional Electronic Evidence	Deepfake Evidence
Authenticity generally indicates reliability.	The authenticity of the file does not establish the truthfulness of its contents.
Hash values and certificates usually support credibility.	Hash values and certificates may validate only the file, not the fabricated content.
Source verification is often sufficient.	Content verification becomes equally important.

In the case Nirmaan Malhotra v. Tushita Kaul, a family court in Delhi was presented with photographs allegedly showing the respondent in an adulterous relationship. The appellant sought to rely on these photographs to challenge alimony awarded by a lower court. The bench observed that in the modern era of deepfakes, the mere production of photographic evidence is insufficient. The burden of proving authenticity shifted to the appellant. The court took judicial notice of deepfake technology’s existence and the ease with which such fabrications could be created.¹³ This judgement, while progressive, exposes the procedural difficulty: how does a defendant prove a negative that a fabricated video is indeed fabricated?

Chain of Custody and the Traceability Problem

Traditional evidence law emphasises chain of custody, the unbroken sequence of custody, control, transfer, and analysis of physical or digital evidence. For deepfakes created with modern algorithms, the “chain” becomes virtually meaningless. A deepfake can be created on an isolated computer, exported as a video file, shared across multiple platforms, and downloaded by law enforcement from social media. At each stage, the “chain” appears intact, yet the evidence is fraudulent at its genesis.¹⁴

Moreover, deepfakes often exploit publicly available information. Creating a deepfake of a public figure requires only sufficient training data: photographs and videos already in the public domain. The technology requires no breach of security, no hacking, and no access to private systems. This creates a peculiar evidentiary problem: the most obvious suspects in a deepfake creation may have no access to any special means, yet prosecution becomes difficult because the “crime scene” is the internet itself, and the “tools” are free, open-source software available to anyone with a computer and basic technical knowledge.¹⁵

Why Chain of Custody Becomes Inadequate for Deepfakes

Traditional Evidence	Deepfake Evidence
The origin of evidence is generally genuine.	Evidence may be fabricated from its very creation.
Chain of custody helps establish reliability.	A flawless chain of custody cannot cure fabricated content.
Evidence is usually tied to a physical or identifiable source.	Creation may occur anonymously using publicly available AI tools.
The investigation focuses on the handling of evidence.	Investigation must also establish whether the content itself is artificially generated.

Major Legal Implications

Existing admissibility standards focus primarily on procedural authenticity rather than factual truthfulness.
Section 65B (now Section 63 of BSA) certification alone may not establish the reliability of AI-generated evidence.
Courts increasingly require independent proof of authenticity where deepfake technology is suspected.
Traditional chain of custody principles are inadequate where fabrication occurs at the point of creation.
Future evidentiary standards may require AI forensic verification in addition to procedural compliance.

Detection, Expertise, and the Black Box Problem

If deepfakes pose authentication challenges, their detection presents equally daunting obstacles. While technological solutions for deepfake detection do exist, including forensic AI tools that analyse facial micro-expressions, detect inconsistencies in lighting and shadows, and identify artefacts in compression patterns, these tools are neither foolproof nor universally accessible to the Indian judiciary.¹⁶

Key Detection Challenges

Authentication of AI-generated evidence remains difficult.
Advanced forensic AI tools are not universally available.
Existing detection technologies are not completely reliable.
Judicial institutions face significant technological limitations.

The Forensic Expertise Gap

For electronic evidence to be properly authenticated under Section 63 of the BSA, an expert must complete Part B of the certification. Yet the legislation provides minimal guidance on who qualifies as an “expert” in the context of AI-generated or AI-manipulated content. The Information Technology Act defines an “Examiner of Electronic Evidence” in Section 79A, yet the qualifications for such examiners remain unclear, and the number of truly qualified professionals in India is vanishingly small.¹⁷

India’s forensic science infrastructure is already strained. According to the Vivekananda International Foundation’s 2025 study on deepfake cybercrime, there are fewer than 150 Central Forensic Science Laboratories across the country serving a population of over 1.4 billion people. These labs are equipped to handle fingerprints, DNA analysis, and basic digital forensics, but deepfake detection requires specialised training in machine learning, GANs, and synthetic media analysis. Most forensic examiners lack this expertise.¹⁸

Current Forensic Capacity in India

Area	Current Position
Authentication Requirement	Expert certification under Section 63 of the BSA
Recognized Authority	Examiner of Electronic Evidence under Section 79A of the IT Act
Qualification Standards	Minimal statutory guidance
Forensic Infrastructure	Fewer than 150 Central Forensic Science Laboratories
Major Challenge	Shortage of experts trained in AI, GANs, and synthetic media analysis

Black-Box AI and Expert Testimony

Furthermore, expert evidence in the Indian system relies on the expert’s ability to explain their reasoning under cross-examination. Yet many deepfake detection tools operate as “black boxes”; they produce outputs (this video is 87% likely to be a deepfake) without clearly explaining the reasoning process. This directly undermines the right to cross-examine the basis of expert opinion under Section 138 of the BSA.

A defence counsel cannot effectively challenge an expert who cannot explain why the algorithm reached its conclusion — the cross-examination becomes a performance, not a test of reliability. In Selvi v. State of Karnataka (2010), the Supreme Court struck down narco-analysis evidence partly because it lacked transparency and reliability. The same reasoning should apply to black-box AI-based deepfake detection tools.¹⁹

Why Black-Box AI Creates Legal Concerns

Lack of transparency in algorithmic decision-making.
Difficulty in explaining forensic conclusions during cross-examination.
Reduced ability of the defence to challenge expert testimony.
Potential conflict with principles of procedural fairness and natural justice.

The “Burden of Proof” Paradox

Perhaps the most vexing question is: who bears the burden of proving that evidence is a deepfake?

Indian criminal law operates under the fundamental principle of ‘ei incumbit probatio qui dicit, non qui negat’ (the burden of proof lies upon the one who affirms, not the one who denies); the burden rests on the prosecution to prove guilt beyond reasonable doubt. If a deepfake is introduced as evidence of the accused’s guilt, and the accused contends it is fabricated, does the burden shift to the defence to prove it is false? Or does the prosecution retain the burden of proving authenticity?

The Nirmaan Malhotra judgement suggests the burden shifted to the appellant. However, this creates an asymmetrical evidentiary problem. The prosecution may introduce a deepfake created weeks or months ago, with no clear chain of custody or origin point. The defence must then somehow prove it did not exist in the form presented at trial. This inverts the traditional burden of proof, particularly in criminal cases where the stakes are incarceration and loss of fundamental rights.

Key Legal Questions

Who should prove whether digital evidence is genuine or manipulated?
Should the burden remain on the prosecution throughout the trial?
Can an accused realistically prove that AI-generated evidence is fabricated?
Does shifting the burden undermine the presumption of innocence?

Burden of Proof Issues at a Glance

Issue	Legal Concern
Introduction of Deepfake Evidence	Authenticity may remain uncertain
Defense Challenge	Difficulty proving fabricated digital content
Chain of Custody	Often incomplete or difficult to establish
Criminal Justice Principle	Risk of reversing the traditional burden of proof

International Approaches: Learning from Global Solutions

While India grapples with deepfake regulation, other jurisdictions have begun charting paths forward, though none have achieved a comprehensive solution. What these approaches offer India is a comparative map of what has been tried, what has worked, and where the gaps remain.

Global Comparison at a Glance

Jurisdiction	Primary Regulatory Focus	Key Strength	Major Limitation
European Union	Transparency and disclosure	Mandatory AI-generated content disclosure	Disclosure can be removed or bypassed
United States	State-level legislation	Victim-focused protections	Lack of national uniformity
United Kingdom	Platform safety and forensic standards	Focus on evidence reliability	Detection tools still evolving
India	Judicial intervention	Protection of personality rights	No comprehensive statutory framework

The European Union: Transparency and Accountability Through the AI Act

The European Union enacted the AI Act (Regulation (EU) 2024/1689), which came into force in June 2024, with key provisions becoming mandatory on August 2, 2026. Article 50 of the AI Act introduces a transparency requirement: providers of generative AI systems must mark outputs in a machine-readable format to make them detectable as artificially generated or manipulated. Deployers of systems generating or manipulating image, audio, or video content – the very definition of deepfakes – must disclose that the content is artificially created. 20

This approach prioritises transparency over prohibition. Rather than criminalising deepfake creation (though harmful deepfakes remain prohibited under separate provisions), the EU mandates disclosure. This allows courts and the public to make informed judgements about synthetic content. The AI Act also imposes significant financial penalties up to 6% of a company’s global turnover for violations, creating powerful incentives for compliance. 21

Strengths of the EU Approach

Mandatory disclosure of AI-generated content.
Machine-readable marking for easier detection.
Strong financial penalties encourage compliance.
Improves transparency for courts, regulators, and the public.

Limitations of the EU Approach

However, the EU approach has limitations. Transparency requirements only help if people actually see the disclosures. Malicious actors can strip metadata or manipulate marking systems. Furthermore, the provisions include exceptions for “evidently artistic, creative, satirical, fictional or similar works”, which could potentially allow deepfakes to proliferate under the guise of satire or entertainment. 22

Metadata can be removed or altered.
Disclosure mechanisms may not always reach viewers.
Creative and satirical exceptions could be misused.

The United States: Fragmented State-Level Responses

The United States has adopted a fragmented approach, with deepfake regulation occurring primarily at the state level rather than through federal legislation. States like Texas, California, and Virginia have criminalised non-consensual intimate deepfakes. Other states target election-related deepfakes. The federal Take It Down Act aims at providing tools for victims to request removal of deepfakes from platforms. 23

From an evidentiary perspective, the Federal Rules of Evidence continue to evolve. Rule 702 allows expert testimony on any subject if the expert’s scientific, technical, or specialised knowledge will help the jury; Rule 901 addresses authentication of evidence. Courts have begun applying these rules to AI-generated evidence, but there is no unified standard. Some courts require forensic analysis to establish authenticity; others rely on expert testimony about the technology’s reliability. The fragmentation creates unpredictability in legal outcomes. 24

Key Features of the U.S. Model

State-specific criminal laws addressing deepfakes.
Special protections against election-related manipulation.
Victim-orientated platform takedown mechanisms.
Evolving judicial standards for AI-generated evidence.

The United Kingdom: Safety-Focused Regulation

The UK Online Safety Act takes a different approach, focusing on the harms caused by deepfakes rather than their technology. The Act places obligations on platforms to identify and mitigate systemic risks, including manipulated media. The UK Forensic Science Regulator has published codes of practice addressing authentication of digital evidence and AI-generated content. Importantly, the UK approach emphasises that forensic examination must be transparent, repeatable, and subject to peer review standards that most current deepfake detection tools do not meet. 25

Highlights of the UK Framework

Focuses on harm reduction instead of banning technology.
Places accountability on online platforms.
Encourages transparent and repeatable forensic practices.
Recognises current limitations of deepfake detection tools.

Common Threads and Critical Gaps

Despite their differences, these international approaches share common threads: transparency requirements, platform accountability, victim protections, and recognition that existing evidentiary frameworks are inadequate. However, all face a common gap: there is no unified international standard for detecting, authenticating, or adjudicating deepfake evidence. A deepfake detected as synthetic in one jurisdiction might be accepted as authentic in another. This creates particular challenges for cross-border cases and international criminal matters.

Shared Global Principles

Transparency obligations.
Platform accountability.
Protection of victims.
Recognition of evidentiary shortcomings.

Remaining Global Challenges

No universal forensic standard.
Cross-border evidentiary inconsistencies.
Different legal thresholds for authenticity.
Growing challenges in international criminal investigations.

Judicial Responses in India: A Cautious Beginning

While comprehensive statutory reform has not yet occurred, Indian courts have begun grappling with deepfake-related matters. In 2025 and 2026, the Delhi High Court and Bombay High Court issued interim injunctions in cases involving unauthorised deepfakes of public figures, particularly celebrities. These orders recognised that personality rights grounded in Article 21 of the Indian Constitution extend to a person’s image, voice, and likeness, and that AI-generated reproductions engage these rights no less than traditional forms of misappropriation. 26

In December 2025, actors including Nandamuri Taraka Rama Rao Jr., R. Madhavan, and Shilpa Shetty secured court orders blocking the spread of deepfake videos and voice clones. The judges held that once notified, intermediary platforms must rapidly remove AI-driven impersonations, regardless of the method used to create them. This represents an important judicial recognition that deepfakes are not merely “new technology” requiring novel rules but a new means of perpetrating age-old harms: misappropriation, defamation, and privacy invasion. 27

However, these civil remedies (injunctions and damages) address only one dimension of the deepfake problem. They do not directly address the admission of deepfakes as evidence in criminal trials, nor do they establish clear forensic standards for authentication. The courts have essentially applied existing doctrines of personality rights and tort law to new technology – a reasonable interim solution but not a permanent framework.

Key Judicial Developments in India

Recognition of personality rights under Article 21.
Protection of image, voice, and likeness against AI misuse.
Rapid takedown obligations for intermediary platforms.
Judicial acknowledgement that deepfakes facilitate traditional legal wrongs.

Continuing Legal Gaps in India

No dedicated deepfake legislation.
No uniform forensic authentication standards.
Limited guidance on admissibility of AI-generated evidence in criminal proceedings.
Existing remedies primarily focus on civil relief rather than evidentiary challenges.

Toward a Solution: Recommendations for Legal Reform

A workable solution to deepfake evidence in Indian courtrooms requires legislative, institutional, and procedural reforms working in concert.

Legislative Reform: A Dedicated Deepfake Framework

India should enact a specific statute addressing deepfakes, distinct from existing sections on forgery, defamation, and cybercrime. Such legislation should:

Define deepfakes clearly, distinguishing between harmful deepfakes (created without consent, intended to deceive or harm) and legitimate synthetic media (clearly labelled as such, created with consent, intended for entertainment or educational purposes).
Establish forensic authentication standards for deepfake evidence, requiring independent verification using internationally recognised detection methodologies, not merely procedural certification under Section 63 of the BSA.
Mandate transparency in AI-generated content, drawing inspiration from the EU AI Act. Platforms, deepfake creators, and content distributors must disclose AI involvement.
Clarify the burden of proof. When deepfake evidence is challenged, the party introducing it bears the initial burden of proving authenticity through forensic means, not merely through procedural compliance.
Create new criminal offences specific to non-consensual deepfakes intended to harm reputation, privacy, or security, with proportionate penalties that reflect the severity of the harm.

Key Legislative Recommendations

Area of Reform	Recommended Measure	Expected Outcome
Definition	Clearly distinguish harmful deepfakes from legitimate synthetic media.	Reduces legal ambiguity.
Authentication	Require internationally recognised forensic verification standards.	Improves evidentiary reliability.
Transparency	Mandate disclosure of AI-generated content.	Enhances public trust and accountability.
Burden of Proof	Require the introducing party to prove authenticity through forensic evidence.	Protects against fabricated digital evidence.
Criminal Liability	Create dedicated offences for malicious non-consensual deepfakes.	Strengthens legal deterrence.

Institutional Development: Building Forensic Capacity

Legislative reform alone is insufficient. India must invest in building forensic capacity to detect and authenticate deepfakes. This requires:

Establishing specialised deepfake detection laboratories within the Central and State Forensic Science Laboratories, staffed with professionals trained in machine learning, synthetic media analysis, and AI forensics.
Creating certification standards for forensic examiners in the field of AI-generated and AI-manipulated media, ensuring that those testifying as experts in court possess genuine, verifiable expertise.
Developing open-source detection tools in coordination with international research bodies, ensuring that deepfake detection methods are transparent, peer-reviewed, and available to all stakeholders—prosecution, defence, and courts.
Training judicial officers in the basics of AI, deepfake technology, and synthetic media. Judges need not become technologists, but they must understand enough to assess expert testimony critically and recognise when a detection tool’s claim of certainty exceeds its actual reliability.

Institutional Capacity Building Priorities

Priority Area	Purpose
Deepfake Detection Laboratories	Strengthen forensic examination of AI-generated media.
Expert Certification	Ensure only qualified forensic professionals testify in court.
Open-Source Detection Tools	Promote transparency and independent verification.
Judicial Training	Improve judicial understanding of AI and synthetic media.

Procedural Reform: Evidentiary Standards for the AI Age

Courts should develop, through case law and rules of practice, evidentiary standards specifically addressing synthetic media. These should include:

A rebuttable presumption of inauthenticity for any digital video or audio evidence when the party against whom it is tendered challenges its authenticity, shifting the burden to the introducing party to prove it through forensic analysis, not merely procedural certification.
Requirements for chain of custody documentation that extend beyond mere transmission data to include analysis of the content itself—forensic examination of compression artefacts, facial inconsistencies, audio patterns, and other markers that expert testimony indicates are present in deepfakes but absent in authentic content.
Mandatory disclosure of detection methodologies used to verify evidence. Experts must disclose the tools used, their margins of error, peer review status, and any known limitations. Black-box AI tools should not be acceptable bases for forensic conclusions.
Corroboration requirements for digital evidence in high-stakes cases. A single deepfake video, even if authenticated, should not be sufficient to convict a person of a serious crime. Corroborating evidence—witness testimony, documentary evidence, forensic evidence—should be required.

Proposed Evidentiary Safeguards

Safeguard	Purpose
Rebuttable Presumption of Inauthenticity	Protects parties against fabricated digital evidence.
Enhanced Chain of Custody	Ensures forensic analysis extends beyond transmission records.
Disclosure of Detection Methodologies	Improves transparency and allows judicial scrutiny.
Corroboration Requirement	Reduces the risk of wrongful convictions based solely on digital media.

Summary of Recommended Reforms

Enact a dedicated legal framework governing deepfakes.
Develop standardised forensic authentication procedures.
Mandate transparency for AI-generated content.
Clarify evidentiary burden when authenticity is disputed.
Create specific criminal offences for malicious deepfakes.
Strengthen forensic laboratories and expert certification.
Train judicial officers in AI and synthetic media.
Adopt AI-specific evidentiary standards through procedural reforms.
Require disclosure of forensic detection methodologies.
Mandate corroborative evidence in serious criminal prosecutions involving digital media.

Conclusion: The Urgency of Action

Deepfakes represent an existential challenge to the evidentiary foundations of Indian jurisprudence. For over 150 years, the Indian Evidence Act served admirably, evolving from paper documents to digital records. Yet the framers of the Bharatiya Sakshya Adhiniyam, despite updating the language and structure of evidence law, did not adequately anticipate a world where digital records could be fabricated so convincingly that forensic analysis itself becomes uncertain.

Deepfakes Challenge Indian Evidence Law

The case that opened this article, the hypothetical trial where a deepfake confession diverts the court’s attention from substantive guilt or innocence to epistemological questions about the nature of evidence, is no longer purely hypothetical. Courts are already encountering these cases. The Rashmika Mandanna deepfake, the Nirmaan Malhotra judgement, and the recent injunctions against celebrity deepfakes – these are early tremors signalling that major legal earthquakes are approaching.

The Path Forward Requires Action at Multiple Levels

The path forward requires action at multiple levels. Legislators must enact dedicated deepfake regulations, drawing insights from international approaches while remaining sensitive to India’s constitutional framework and the imperative to protect fundamental rights. Institutions must develop the forensic and technical capacity to authenticate synthetic media with reliability and transparency. Courts must evolve their evidentiary doctrines, recognising that procedural compliance is distinct from substantive authenticity. Legal professionals must educate themselves on AI and synthetic media, recognising that ignorance of these technologies is no longer an acceptable posture in a modern courtroom.

Enact dedicated deepfake regulations.
Strengthen forensic and technical authentication capabilities.
Refine evidentiary doctrines to distinguish admissibility from authenticity.
Promote AI and synthetic media education for judges, lawyers, and investigators.

Why Authentic Evidence Matters

Most critically, we must remember why evidence law exists: to advance justice. When technology undermines the reliability of evidence, it undermines the possibility of justice itself. The accused cannot receive a fair trial if fabricated evidence is indistinguishable from authentic evidence. The victim cannot receive justice if real evidence is dismissed as a deepfake. Society cannot maintain rule of law if courts cannot distinguish truth from falsehood. The urgency, therefore, is not academic; it is profound and existential.

Challenge	Impact on the Justice System
Fabricated digital evidence	Threatens fair trials and evidentiary reliability.
Weak authentication mechanisms	Reduces confidence in digital records.
Lack of dedicated legal framework	Creates uncertainty in judicial decision-making.
Limited AI awareness among legal professionals	Increases the risk of wrongful reliance on synthetic media.

The Future of Indian Courts in the Age of Synthetic Media

As India’s judicial system enters the era of synthetic media, it must do so with clear eyes, clear laws, institutional capacity, and procedural sophistication. The alternative allowing deepfakes to infiltrate courtrooms unchecked risks transforming trials from quests for truth into theatrical performances where the most convincing narrative, regardless of authenticity, prevails. That would be a betrayal of the promise of justice itself.

References and Citations

Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., … & Bengio, Y. (2020). Generative Adversarial Networks. Communications of the ACM, 63(11), 139–144.
Mirsky, Y., & Lee, W. (2021). The Creation and Detection of Deepfakes: A Survey. ACM Computing Surveys, 54(1), 1–41.
Polilegals.com (2025). Deepfake Technology in India: Navigating Legal, Ethical, and Societal Implications.
URL: https://polilegals.com
Bharatiya Sakshya Adhiniyam, 2023 (India), Section 63(4)(c).
LiveLaw (2024). Understanding E-Evidence Under Bharatiya Sakshya Adhiniyam 2023: Key Provisions and Implications.
URL: https://www.livelaw.in
International Journal for Multidisciplinary Research (2025). Deepfake Evidence and the Indian Criminal Justice System.
Information Technology Act, 2000 (India), Sections 66C and 66D.
Information Technology Act, 2000 (India), Section 66E.
National Stock Exchange of India Ltd. v. Meta Platforms Inc. & Ors. (2025). Delhi High Court.
The Criminal Law Blog (2025). The Authenticity Challenge: Addressing the Concern of Producing Deepfake-Generated Media as Evidence in Courts.
Anvar P.V. v. P.K. Basheer, (2014) 10 SCC 473.
Arjun Panditrao Khotkar v. Kailash Kushanrao Gorantyal, (2020) 7 SCC 1.
Nirmaan Malhotra v. Tushita Kaul (2024). Delhi High Court.
IJMR (2025). Deepfake Evidence and the Indian Criminal Justice System. International Journal for Multidisciplinary Research.
Thomson Reuters (2025). Deepfakes on Trial: How Judges are Navigating AI Evidence Authentication.
Agarwal, A. (2025). Deepfakes and Indian Law: An Urgent Need for Regulation. Indian Journal of Law and Technology, 17.
Information Technology Act, 2000 (India), Section 79A.
Vivekananda International Foundation (2025). Bharatiya Laws Against Deepfake Cybercrime: Opportunities and Challenges.
Selvi v. State of Karnataka, (2010) 7 SCC 263.
EU AI Act (Regulation (EU) 2024/1689), Article 50.
Wiggin LLP (2026). EU AI Act and Deep Fakes: Towards a Future Code of Practice on AI Output Transparency.
Ondato (2026). Deepfake Laws Explained: Global Regulations and Legal Risks.
IJERT (2025). Deepfake Cyber Threats in India: An Emerging Challenge Without Legal and Technical Safeguards.
Global Law Experts (2026). Deepfake Injunctions India – Latest High Court Orders.
UK Forensic Science Regulator (2023). Code of Practice for Digital Evidence and Synthetic Media. Home Office.
Constitution of India, Article 21 – Right to Life and Personal Liberty.
National Law Review (2026). Lights, Camera, AI, Action: India’s Recent Celebrity Deepfake Lawsuits.
Document Prepared: June 24, 2026 | Suitable for legal education purposes | Reference material for legal professionals and practitioners.

What's Hot

Tags

Categories