INTRODUCTION
With the ability to make a video of someone saying things that they didn’t, the issue before the Indian courts is not just philosophical anymore; it is a pressing practical question. In this digital age, the use of deepfake – AI-generated synthetic media – has been getting increasingly sophisticated and is directly threatening the pillar of trust in the evidence that forms the backbone of the Indian legal system.[1]
With a growing number of courts responding to electronic evidence, the concern of fabricated AV material is an ever-present danger. In this blog, I explore the intersection of “artificial reality” and “judicial truth,” and question whether the Indian evidence law, largely developed in an analogue era, could identify, exclude, or place in context evidence created with the help of deep fakes.
UNDERSTANDING DEEPFAKES: TECHNOLOGY BEHIND THE DECEPTION
Deepfakes are created with Generative Adversarial Networks (GANs), a type of machine learning that consists of two neural networks: a generator and a discriminator. Around 2017, it gained popularity as face-swapping videos were created using AI and posted on social media.[2] Since then, technology has progressed from basic digital fake to hyper-realistic synthetic video, capable of mimicking voice, micro-expressions, and mannerisms of the human face, and more, with almost the same level of fidelity.
In the Indian context, deepfakes have already proved to be a disruptive force, as they have been used to create fake videos of politicians and public figures to disrupt elections and to make intimate images of women without their consent to harass them. The next step in this threat becoming evidence in court is the next new challenge—and the legal system is clearly not ready for it.
THE EVIDENTIARY FRAMEWORK: DOES INDIAN LAW COPE?
The Indian Evidence Act 1872 (with its amendments) and the Information Technology Act 2000 are the two major legislations governing the admissibility of electronic evidence in India. The conditions for the admissibility of electronic records are prescribed under Section 65B of the Evidence Act, which requires a certificate of integrity and authenticity of the device and of the record.[3]
In Arjun Panditrao Khotkar vs Kailash Kushanrao Gorantyal[4]The Supreme Court of India had stated that a Section 65B certificate is mandatory, and electronic evidence without such a certificate is not admissible. This can be optimised to exclude unverified electronic records; however, it was not originally written with the inclusion of deepfakes. In principle, a malicious actor could present a deep-fake video with a fake Section 65B certificate from an untrusted or compromised certifier and introduce fake evidence that satisfies the formal admissibility threshold, but does not actually constitute evidence.
The Bharatiya Sakshya Adhiniyam 2023 (India’s Evidence Law) also has provisions pertaining to electronic records, though these are quite similar; there is no specific provision in place to address synthetic media or AI-generated evidence.[5]. In a world as fast-changing as deepfake technology, it’s hard to imagine that nothing has been done in the legislature.
JUDICIAL CHALLENGES: DETECTION, RELIABILITY, AND EXPERTISE
The threshold problem is one of detection. The court does not have AI forensic capabilities or tools — nor should it! To identify deepfakes, advanced analytical capabilities, such as frame-level inconsistency analysis, biometric liveness detection, and metadata analysis, are needed.[6] India is still facing a problem of not having a standardised judicial process or a government-approved forensic lab that can certify the authenticity of media created using AI.
Deepfake forensics has not been specifically discussed in a Law Commission of India report. In contrast, other jurisdictions, like the United States, have had legislative changes, with California’s AB 602 and AB 730[7] Specifically addressing deepfake misuse in legal and electoral settings. The AI Act 2024 introduces transparency requirements for synthetic content created by AI, such as watermarking and disclosure of the content’s origin.
In the event of an occurrence of deepfake evidence, the experts who provided the evidence would be the first field to contest it, in accordance with the rules of the Bharatiya Sakshya Adhiniyam 2023 (Indian Evidence Act 2023). But the lack of qualified digital forensic experts in India, along with the lack of standards for AI forensics certification, makes this path somewhat uncertain.[8]
CRIMINAL MISUSE AND PROSECUTORIAL DIMENSIONS
The potential ways deepfakes could be used in a crime case could be many: A fabricated confession, a manufactured alibi video, a fake security camera footage, and a fake witness testimony. In each scenario, the epistemological heart of adjudication (the court’s power to establish the truth from the record) is struck.[9]
While Section 67A of the IT Act is applicable to the publication of obscene and sexually explicit material, and Section 66C is applicable to identity theft, both these sections only partly address the misuse of deepfakes. But there is no independent crime for making or using a deepfake for the purpose of fooling a court.[10] The Indian Penal Code (Bharatiya Nyaya Sanhita 2023- henceforth BNS) has provisions to punish the fabrication of false evidence (Section 192 BNS), but they were formulated before the era of synthetic media.[11]
In Virender Kumar v State of Himachal Pradesh[12]The High Court warned against relying on the electronic evidence without any critical scrutiny, emphasising that it must be corroborated and the authenticity of the evidence must be established. This instinctive power of the judge, which is valid, is still an ad hoc measure and not part of a system.
REFORM IMPERATIVES: A WAY FORWARD
There is a need for action on several fronts to realise meaningful reform. First, the legislature should set out a clear evidentiary standard for AI-produced or AI-altered media that mandates disclosure of synthetic media and allows the media to be better authenticated.[13] Second, policies and procedures for the detection of Deep Fakes should be jointly developed by the Ministry of Electronics and Information Technology and the National Forensic Sciences University, to establish a pool of experts that can be accepted by courts.[14]
Third, courts should adopt a rebuttable presumption framework: If the party pointing out the possible fabrication of evidence has credible evidence, then the burden should be placed on that party to prove its authenticity by using other independent forensic means. Fourth, India should be a part of the international frameworks, such as requirements for provenance and watermarking in the EU Act on AI, as a template for India’s own AI governance.[15]
Deepfakes have already been raised as a national issue by the Press Information Bureau and the Election Commission during election seasons. The obvious next step is to extend the regulatory focus to judicial proceedings, and it’s needed.
CONCLUSION
Judicial proceedings are based on evidence, and evidence must be reliable. The so-called “deepfakes” are designed to do just this. The existing Indian evidence law is structurally inadequate to deal with this challenge. But the fact that deepfake technology is so advanced compared to legal protections is no simple matter of technology, just as it isn’t a matter of technology that the game is coming to Australia. It’s not just an inconvenience that deepfake technology is so advanced compared to legal protections; it’s a threat to the rule of law. Idealistic changes in the law, capacity building for the forensics, and judicial sensitivity will not be luxuries; they will be essentials. A courtroom shouldn’t be another place of artificial reality.
References:
[1] Nina Schick, Deepfakes: The Coming Infocalypse (Twelve 2020) 1–14.
[2] Ian Goodfellow and others, ‘Generative Adversarial Nets’ (2014) 27 Advances in Neural Information Processing Systems 2672.
[3] Indian Evidence Act 1872, s 65B
[4] Arjun Panditrao Khotkar v Kailash Kushanrao Gorantyal (2020) 7 SCC 1.
[5] Bharatiya Sakshya Adhiniyam 2023, ss 61–63
[6] Umur Aybars Ciftci, Ilke Demir and Lijun Yin, ‘FakeCatcher: Detection of Synthetic Portrait Videos Using Biological Signals’ (2020) IEEE Transactions on Pattern Analysis and Machine Intelligence.
[7] California Assembly Bill 602 (2019)
[8] Indian Evidence Act 1872, s 45; Bharatiya Sakshya Adhiniyam 2023, s 39
[9]Robert Chesney and Danielle Citron, ‘Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security’ (2019) 107 California Law Review 1753, 1788–1800.
[10] Information Technology Act 2000, ss 66C, 67A.
[11] Bharatiya Nyaya Sanhita 2023, s 228 (fabricating false evidence)
[12] Virender Kumar v State of Himachal Pradesh (2019) HPHC.
[13] MeitY, ‘National Strategy on AI’ (NASSCOM 2018) 31.
[14] National Forensic Sciences University Act 2020
[15] Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence [2024] OJ L 1689/1, art 50 (transparency obligations for AI-generated content).