Criminalizing Deepfakes - Legal Challenges And The Future Of Cybercrime Law In India

Author: LawSphereX

Category: Criminal Law

Published: 2026-08-23T19:03:57.096Z

Updated: 2026-08-30T07:11:11.885Z

CRIMINALIZING DEEPFAKES: LEGAL

CHALLENGES

AND THE FUTURE OF CYBERCRIME LAW IN INDIA Mahi Soni, BA.LLB (Hons), Manipal University Jaipur, Rajasthan, Dahme Kalan, India

ABSTRACT

Deepfake, a hyper-realistic synthetic medium produced by artificial intelligence (AI), has developed into a potent and dangerous tool for distorting reality in unprecedented ways in the digital age. By manipulating videos, audio, and images to convincingly show individuals saying or doing things (they did not) this strongly threatens individual privacy, integrity, political and social stability, and the possible administration of justice. Deepfake technology is evolving rapidly, and AI tools are more readily available. Legal and ethical challenges, mostly as a result of technology evolving faster than the law, are arduous—more so in jurisdictions like India where the law is still developing to catch up with technological advances.

Heeding this context, in this paper I critically examine India’s existing cyber and criminal law framework, while noting how there is currently no explicit legal recognition or regulation of deepfakes. While there may be some provisions that can address instances of deepfakes under the Information Technology Act, 2000 And Provisions of BNS, 2023 such as defamation, obscenity, or violations of privacy—there is still no legal clarity or prohibition regarding the more complex issues, such as AI-driven impersonation, digitally manipulated consent, or extensive disinformation in politics or society. In addition, there are other constitutional implications of regulating deepfakes, principally the right to freedom of speech and expression, versus not infringing someone's right to privacy, dignity, and democratic integrity, etc.

Also, the evidentiary challenges of determining the source and/or the intent of deepfakes make enforcement more complicated. At the end of the paper, I suggest a more proactive regulatory situation where the law is update to technology and holds citizens in an ethical space. Keywords: Deepfake, Crime, Misuse of AI, face manipulations

INTRODUCTION

The emergence of deepfake technology, which utilizes artificial intelligence (AI) and machine learning, has initiated a phase of digital content alteration. Deepfakes are regarded as hyperrealistic synthetic media, or a form of altered media (including videos, audio, or images), which will show a single subject doing or saying something they did not actually do, i. e. , capable of changing the perception of the event. Under ideal circumstances, deepfakes may have many positive applications by enhancing creativity and media literacy. However, misapplication of deepfake technology mindsets poses significant threats with longstanding negative implications for personal privacy, reputation, national security, and democratic integrity. “In India, the proliferation of deepfakes has led to several alarming incidents.

For instance, a man in Navsari, Gujarat, was arrested for sharing a deepfake video of Prime Minister Narendra Modi in a WhatsApp group. The video, which morphed the Prime Minister's voice and image, contained disparaging remarks about Operation Sindhoor and the Indian Army, aiming to spread fear and misinformation. Such incidents underscore the potential of deepfakes to disrupt public order and national security”. 1 In a more technical sense, deepfakes (short for Methods are utilized to create "deep learning" and "fake. ") which can superimpose face images of a target person onto a source individual's video to produce a video of the target person doing or saying things the source person does. This is one of the deepfakes, that is, face swap. In a more general definition, deepfakes are synthetic intelligence-created content that is also classified under two other categories, i. e.

, lipsync and puppet-master. Lip-sync deepfakes are videos that have been altered to render the mouth movements in sync with an audio deepfakes. Puppet-master deepfakes consist of recordings of a target object (puppet) which is triggered after the facial expression, head and eye movement of the other One person (master) standing in front of a camera. Also deepfake require large amount of photos and videos which are generally of any celeb or politicians etc. to harm their reputation or even for the fun purpose sometime people create porn videos of the female celebrities (“The first deepfake video emerged in 2017 where face

1 https://timesofindia.indiatimes.com/

of a celebrity was swapped to the face of a porn actor)”. for the fun and view and this is crime then also sometime we can’t do anything because of dangerous protected AI. Deepfakes thus can be: “abused to cause political or religion tensions between countries, to fool public and affect results in election campaigns, or create chaos in financial markets by creating fake news. It can be even used to generate fake satellite images of the Earth to contain objects that do not really exist to confuse military analysts, e.g., creating a fake bridge across a river although there is no such a bridge in reality. This can mislead a troop who have been guided to cross the bridge in a battle.”2

NATURE AND

SCOPE

OF DEEPFAKES The term "deepfake" comes from "deep learning" (a subgroup of AI) and "fake." Deepfakes are hyper-realistic synthetic media (i.e., videos, audios, and images) created by using Generative Adversarial Networks (GANs), a type of machine learning framework. In a GAN, two separate AI models (generator and discriminator) challenge each other progressively producing more realistic outputs. Deep fakes can reliably manipulate face, voice, and body language to create an appearance of someone saying or doing something they never did. “According to Chesney and Citron (2019), deepfakes threaten to erode trust in both individual identity and objective truth, potentially leading to a “reality apathy” where people no longer trust visual media at all. This raises serious concerns for democracy, criminal justice, reputation, privacy, and public safety”3.

2.1 EVOLVING USE-CASES AND HARMS

A. Political Deepfakes and Misinformation “Deepfakes have been increasingly weaponized in politics. In January 2024, the Times of India reported that a deepfake video of Prime Minister Narendra Modi, falsely depicting him making inflammatory statements, went viral during Gujarat election season. This video was widely

2 Patrick Tucker. The newest AI-enabled weapon

‘deep-faking’ photos of the earth. https://www.defenseone.com/ technology/2019/03/next- phase- ai- deep- fakingwhole-world-and-china-ahead/155944/, March 2019.

3 Chesney, R., & Citron, D. (2019). Deepfakes and the New Disinformation War, California Law Review

shared across WhatsApp and social media platforms before authorities confirmed it was fake”. Such political deepfakes can manipulate voter behaviour, create communal tension, and undermine democratic processes. “Additionally, in the 2020 Delhi Assembly elections, deepfakes of BJP leader Manoj Tiwari were circulated in multiple languages to reach diverse voter bases. While in that case the content was campaign material created by the party itself, it raised concerns over disclosure and ethical boundaries of synthetic political messaging”4. B. Gendered Harassment and Deepfake Pornography The creation of non-consensual pornography is one of the most common and harmful uses of deepfake technology. Celebrities Rashmika Mandanna and Emma Watson have suffered as victims of deepfake adult content on social media platforms.

Deepfake pornography generates synthetic videos that merge real faces onto pornographic bodies using artificial intelligence layering. These uses violate a person's privacy, dignity, and bodily autonomy. In India, it is common for this type of content to be posted on Telegram or Instagram without any content moderation. Victims of this type of content will face significant emotional distress and limited options for legal action due to a lack of a specific legal provision regarding the making or sharing of deepfake content. C. Financial Fraud and Corporate Deepfakes “Deepfakes are also being used in white-collar crime. In 2023, a multinational firm in Hong Kong reported a case where an employee transferred $25 million after attending a Zoom call with what appeared to be the company’s CFO and executives — all of whom were actually deepfakes created using AI voice and video tools”5.

In India, similar impersonation risks exist in corporate contexts, especially the telecommunications and fintech arena where voice-based authentication is frequently adopted. This type of AI-augmented fraud sits between cybercrime and economic crimes, to some extent.

4 Hindustan Times, February 2020. Delhi elections

Deepfake videos of BJP’s Manoj Tiwari go viral.

5 CNN, Feb 2024. Employee duped by AI deepfake video into wiring $25 million

2.2 LEGAL AND CONSTITUTIONAL IMPACTS

Deepfakes infringe upon the following constitutional rights:

1 Right to Privacy

In Justice K.S. Puttswamy v. Union of India, (2017)6, the Constitutional Court recognized the right to Privacy as fundamental under Article 21 of the Constitution, which grants an individual autonomy and control over information and privacy, or informational privacy (Venkatesh, 2013) around, for example, misleading a facsimile representation, through a fraud or recklessness that informs speech and implication.

2 Right to Reputation and Dignity

The Supreme Court in Subramanian Swamy v. Union of India (20167) held the right to reputation as falling under the right to life, see Article

21. In contexts where deepfake pornography and impersonation can destroy a person's

public reputation, dignity is lost, and there may never be repair to reputational harm.

3 Right to Free Speech and Expression (Article 19(1)(a))

While this right to speech includes satire and dissent in the political context, deepfakes that distort the nature and facts of relationships with the intent to promulgate misinformation and harm, may not be reasonable expression from the point of view of the communications as implications and inferences derived and designed.

2.3 SCOPE

OF THREAT IN THE INDIAN CONTEXT The threat of deepfakes in India is not merely potential; it is actual, pressing, and developing at an alarming pace. The Indian context is particularly conducive to synthetic media abuse due to many factors: a large user base with limited digital literacy, a highly polarized political situation, languages that make verification difficult, and a legal framework that has not caught up with technological advancements. Digital Penetration and Information Vulnerability. India has one of the largest internet user bases globally, with over 850 million users, many of whom obtain information via mobile phones and social media platforms like WhatsApp,

6 (2017) 10 SCC 1

7 (2016) 7 SCC 221

Facebook, and Instagram. The 2023 Digital News Report by the Reuters Institute suggests that a large slice of India's population consumes news through forwarded messages and short video clips often without questioning whether it might be false. In a virtual environment hostile to and prone to misinformation, deepfakes particularly in vernaculars can be misconstrued and misidentified as real content, creating a higher risk for being duped by untruthful content. Emerging Cases of Political Misuse. Evidence of the use of deep fakes for political deception has already emerged in India. In January of 2024, a deepfake video involving Prime Minister Narendra Modi making disparaging comments surrounding the Indian Army went viral in Navsari, Gujarat in the leadup to local elections.

Subsequently, the Gujarat Police arrested the alleged perpetrator invoking Section 153A of the Indian Penal Code 1860 which prohibits the promotion of enmity between groups, and Section 505(1)(b) of the Indian Penal Code 1860, which deals with the formation of public alarm. The case brought to light the possibility of much more effective extortion of public order and implications for the outcomes of elections by using synthetic content as a weapon.

LEGAL FRAMEWORK

FRAGMENTED AND INSUFFICIENT India's legal accountability to deepfakes is fragmented and circuitous. The Information Technology Act, 2000, currently the leading law on cyber offence recommendations and, on its full statutory range, is limited: ITA, 2000 Section 66D- penalises cheating by impersonation with computer resource; however, applicability to deepfakes is limited and requires proof of intent to deceit for wrongful gain. ITA, 2000 Section 67- deals with the publication of obscene content in electronic form; while it may provide some base to include specific forms of deepfake pornography; it does not extend to non-consensual synthetic media acts, which do not either contain obscenity or act in breach of dignity. The newly passed Bhartiya Nyaya Sanhita, 2023 contains provisions equivalent to Chapter 4B of the Indian Penal Code which include:

BNS, 2023 Section 69- criminalizing impersonation by electronic means; • BNS, 2023 Section 73- penalising publication of false or misleading information to cause public mischief or alarm. Despite these additions, there is no distinct legal definition or offence relating to the act of creating or sharing deepfakes, resulting in obscurity and under-enforcement.

EVOLUTION OF DEEPFAKE GENERATION TECHNIQUE This section will address the history and present-day deepfake-generating methods, and we will discuss the important role that artificial intelligence (AI) plays as a dominant technology. This study has gathered evidence it must present about the complicated technological processes necessary for the construction of a deepfake, and although this document examines these processes, it prepares you in a methodical fashion for a complete understanding of the datasets that the computer scientists, software engineers and machine learning engineers will need to reference and the necessary datasets to develop effective detection methods, in order to connect theory to practice through analysing the concept of using deepfakes.

As there are new data creation, data mining, and data manipulation technologies, computer graphics have been using and supporting video modification technologies using means for decades. They often utilise a new reconstructed version of the videos face geometry in the new videos that they create. There are two subjects of note here in the deepfake generation research field - generation methods and datasets. Notably, deepfakes are a technology based upon generative adversarial networks. Like many technologies for specific social purposes, most research teams engaged on the development of GANs are focusing upon ameliorating levels of image and video quality in adopting a public-based perspective. There is also evidence that GAN-based synthesis strategies can generate surprisingly high-quality forged videos.

The indications are strong from the research that video and image manipulation is growing due in part to the technological changes in machine learning and deep learning.

3.1 APPLICATION OF ARTIFICIAL INTELLIGENCE

Deepfake relies on artificial intelligence (AI) to produce or change visual and audio content with a high potential to mislead. GAN framework was made by "Goodfellow (2020)"8 in 2014. Several researchers have explored computer vision methods from work that has been connected to the process of creating deepfakes, that made use of different neural network models and architectures, mainly GAN. GAN is a tool to create fake videos or images GAN has had a number of changes and improvements since first introduced in 2014. It's now become easier to use a pre-trained GAN to swap a face of a person in an image or video with the face of a different person in a matter of moments. Additionally, GUI based applications, such as Fake App, have made it once again easier for laypersons to create these deepfakes.

Technology is so advanced that, now, individuals with a strong desire, time, and computing power can use it.

3.2 CATEGORIES OF TECHNIQUES FOR FAKE MEDIA GENERATION

Techniques for generating fake media can be split into traditional techniques and deep learningbased techniques. Both of these techniques need media editing tools to produce convincing fake media. Traditional fake images and fake videos generation, typically utilize computer vision and image processing methods. As these before deep learning methods have been developed before deep learning, the generative images would not be as realistic as those made using deep learning methods. We classify the traditional fake image generation methods into four types based on the intended use: total face synthesis, attribute manipulation, identity swap, and expression swap/face re-enactments. A. ENTIRE FACE SYNTHESIS: this means that the entire face is swapped or morphed from another face in short exchange of face with one another. B.

ATTRIBUTE MANIPULATION: This means certain type of modification or manipulation of image or content known as deepfake of the image to create the realistic fake. C. IDENTIFYING SWAP: This approach can be used to modify a video or image by placing one individual’s face onto another individual’s body. CNN to develop a faceswapping model. 8 Kaur, A. , Noori Hoshyar, A. , Saikrishna, V. et al. Deepfake video detection: challenges and opportunities. Artif Intell Rev 57, 159 (2024). https://doi. org/10. 1007/s10462-024-10810-6 CONVERSATIONS AND FUTURE RESEARCH DIRECTIONS With the help of deep learning, deepfakes can be created faster than ever before, and with the emergence of social media, the dissemination of these fake contents is also faster. Based on my observation, for the purposes of sabotage, those who create deepfake content with malicious intent do not need to reach a wide audience.

They deliver it to their designated target audience without media awareness. This could be the case with intelligence organizations trying to garner an outcome that alters the decisions of influential people, such as politicians, which leads to threats to national and international security. The alert went out about the deepfake danger and the research community went to work to create deepfake detection algorithms and numerous findings have been reported. The review of state-of-the-art techniques of this paper is an overview of an assortment of measuring devices. Employing detection techniques to locate deepfakes is essential, but grasping the real cause of individuals sharing the significance of deepfakes is magnified.

That aspect is important because deepfakes are progressively attaining a greater level of photorealism, and it is considerably anticipated that detection software will be lagging behind deepfake creation technology. A study on social deepfakes context to help users make such judgement is therefore worth undertaking Videos and photographics have commonly been utilised as evidences in police investigation and justice cases. They can be submitted as evidences in a court of law by digital media forensics with a background in computer or law enforcement and computer skills in the acquisition, investigation, and analysis of computer information. The artificial intelligence and machine learning technology advancement could have been used to change such electronic materials.

Therefore, specialists' opinions may not be enough for verification of such evidence, as even experts find it challenging to identify tampered materials. This aspect calls for in present-day courts, it is necessary to consider the implications of using images and videos as evidence in convicting criminals, the fact that there is a broad range of digital manipulation techniques.

CONCLUSION

Deepfakes have already undermined confidence of individuals in media content because viewing them is no longer synonymous with believing them. They can harm and traumatize the targeted individuals, increase the rate of hate speech and disinformation, and even might generate political tensions, incite the public, violence or war. It is particularly vital now because technologies on how to develop deepfakes are becoming more readily available and social media can disseminate those false contents very rapidly. This survey gives an overview of state-of-the-art deep fake generation and detection techniques and gives a general overview of challenges, possible trends, and future directions in this field. This work will thus be of use to the artificial intelligence research community to formulate effective strategies to combat deepfakes.