Introduction

Introduction of Artificial Intelligence (AI) and its development into a deeply refined system which results in realistic outcomes has brought forth an epidemic of misinformation, privacy concerns and defamation issues. The Deepfakes are essentially either cured content or content entirely created from scratch in the form of image, audio or video. As the name in itself suggests, it is a content not real in nature. It could be used to make it appear someone said or did something they never did in real life.^[i] Hence, an alternate reality is created and disseminated in the digital world, which further affects the life of the person involved in such content. With the quick access nature of this AI system, no one is spared from becoming the victim of deepfake.

With the UN addressing the issue of rampant misuse of AI to create deepfakes^[ii] and focusing on the aggravating consequences that women are facing with defamatory images and videos^[iii]; and explicit fake content being generated of artists^[iv] and ordinary public, this creates the basic concern of privacy breach and harm to the reputation. 

To understand how AI generates such believable content, it is firstly to learn that the Deepfakes are created with deep learning models and primarily Generative Adversarial Networks (GANs). This model is structured to develop itself to replicate human likeness through training on large datasets of images and videos. The system encodes and decodes facial features by analysing the photographs and videos, and then map the individual’s expressions onto another’s identity, producing highly realistic synthetic media as content. It is dependent upon the availability of data, which is ubiquitous, and software, which is becoming increasingly accessible, thus facilitating the technology for legitimate use as well as for extensive misuse, especially in the production of non-consensual intimate images.^[v]

India and the U.S.- The Impact of Deepfakes on Society and Women

The crux of the deepfakes of women in all spectrums of the society lies with the societal mentality of inequality, a pre-existing scenario of online harassment of women, and objectification of women. This fuels the misuse of AI tools to create deepfakes which project non-consensual intimate imagery to further feed on to the sexual exploitation in the society. The feasibility of the technology has made it so accessible to deploy AI for creating such explicit contents on women without even getting traced or being held liable easily. The damage of such acts are, on the other hand, detrimental to the reputation of the person involved in the content and further causing deep psychological distress to them. In case of any minor feud with a woman, someone could use such deepfakes as a mark of vendetta.

In India’s case, current reports emphasize the point that the rates of deepfake-based abuse of women are rising at a rapid and alarming pace. As reported by The Economic Times, the number of women victims of deepfake-based abuse has escalated from 93 to over 900 in a short period of time.^[vi] This is an indication of the exponential rise of such crimes. The rise of such crimes is due to the availability of generative AI and the lack of effective deterrents for such abuse.

India is currently dealing with both the celebrity and the public targets in case of such deepfakes. The recent news on morphing actress Rashmika Mandanna’s likeness into inappropriate content in 2023^[vii] demonstrates the ease with which the publicly available images can be distorted and misused. Similar situations have been observed in a number of other cases, like in Aaradhya Bachchan v YouTube Channels^[viii], a case decided by the Delhi High Court in 2023, the judiciary stepped in to check the spread of misleading and damaging information involving a minor, realizing the damaging effects on reputation and psychological well-being. While morphing and distortion in deepfake is subjected to male public figures and the general male community as well, but the frequency of cases is not much higher in females. Especially, looking at the case involving Aradhya, even a child is roped into such a serious level of misuse. In Kamya Buch v JIX5A^[ix], decided by the Delhi High Court in 2025, relief was granted against defamatory content generated through AI, suggesting a recognition of emerging harms by the judiciary. These cases, therefore, indicate how, even as public figures are obvious victims, the tech behind them may be just as, if not more, damaging for regular women without access to legal recourse. The gravity needs to be acknowledged of such incidents, where a woman, functioning in a society shall be subjected to an intense level of mental trauma because of dissemination of such content. Not only the right to privacy is getting affected but the very basic right to live with dignity is impacted underArticle 21 of the Constitution of India^[x].

The U.S. reflects a 2026 report by the United Nations News, which points out that there are systemic failures when it comes to protecting women from AI-based exploitation. In the report, it is emphasized that victims of deepfake abuse are faced with jurisdictional hurdles and difficulties in providing evidence of abuse.^[xi] These points indicate that there has been a shift from individual deepfake abuse to a systemic form of digital abuse.

In the United States, the impact on women becomes similar to the condition sustaining in India. With the widespread circulation of deepfake images of pop singerTaylor Swift in 2024^[xii], the depiction of public figures also being the targets to such reputational misuse. In terms of the legal perspective, the case of Thaler v. Perlmutter^[xiii] provides clarity regarding the legal position of AI-produced material and brings in the concern of the need for the framework of liability. The exploitative and capitalist structure of the nation enables the instances like Grok/xAI class action suit of 2026^[xiv],where platforms like ‘X’ are used to create deepfake content and procure profit out of such non-consensual explicit content.It shows how the relationship between technological advancement and gender-based harm is not coincidental but rather systematic in nature.

Impact on Public

Data has repeatedly reflected that women have a higher tendency to be the targeted victims of deepfakes used in pornography sites or other social spaces, without consent. According to studies, 90-95% of the total deepfake videos circulating on the internet contain non-consensual sexual material, and the majority of these victims are women.^[xv] The psychological impact of deepfakes on women ranges from anxiety and damage to their reputation and professional life to long-term mental trauma, which is far more detrimental than other forms of harassment due to the permanent and widespread nature of digital media. It is to be observed carefully that the aftermath of such nature of digital reach extends not only to the reputed and prominent individuals, but to all citizens on a day-to-day basis, including professionals, minors and regular people. In the case of the U.S. of ‘Lancaster Country Day School’ of 2024, there was a spread of AI-contorted explicit images of students. This shows how ordinary citizens are being targeted. This magnifies the harm and simultaneously also reduces one’s ability to control their online presence.

From a societal perspective, the misuse and constant exploitation of the audiovisual content especially has a detrimental impact on women’s engagement online. Thus, this promotes further base for gender inequality to flourish. Such abuse of content creates a sceptical point of view towards all the audiovisual content in general, which affects not only the individual but also information systems as a whole which presents the content, and their business.^[xvi]

Analysis of Laws

Over the spread of defamation laws present, laws of Privacy, harassment and the angle on the tort law, the criminal provisions and the platform liability frameworks, it is intended to explore and study all the stated aspects present in India. When looking at the current scenario of the Indian Laws and provisions towards combating the legal issues cropping from deepfake, it is still spread out across a combination of criminal law, tort laws, cyber laws, and other relevant laws.

Section 356 of the Bharatiya Nyaya Sanhita (BNS)^[xvii] deals with penalty for defamation, and imposes an imprisonment for up to two years, a fine or both for causing reputational harm under criminal liability. Although, observing its usage in deepfake cases, the application is rather seldom. The provision recognises identifiable authorship and the presence of intention, both being elements that are difficult to prove in case of deepfake content as the anonymity exists in its dissemination across multiple social networking platforms. Hence the attribution issue crops up. Moreover, the requirement of a high evidentiary threshold becomes a difficult task on acts involving AI based media. Civil defamation, under tort law, enables victims to claim damages. Yet it becomes a cumbersome process given its time taking nature. The instant nature of dissemination of the deepfakes makes it crucial for the remedy of such acts to be quick as well. Especially, in case of women, where the reputational damage is quickly sexualised and socially stigmatised. While the legal framework provides protection against defamation, harassment, and privacy infringement to a certain level, in totality, the structure still remains inadequate in terms of defining the extent, anonymity, and technology used in deep fake abuse, which fades out even more when focusing on crimes against women through AI deepfakes.

The Section 66E of the Information Technology Act 2000^[xviii] which deals with privacy laws, it states that any capture, publication, or transmission of images without consent and being a deliberate act shall attract it being criminalised. And although the provision can be read in terms of addressing the deepfake content, the scope of this particular provision is clearly restrictive since it specifically limits itself in its interpretation to the ‘real’ images, which in turn excludes the synthetic or AI-generated content. Adding to the issue, the images that are morphed in deepfakes are usually publicly accessible, cancelling the ‘private’ nature of the image which is addressed by the provision. And this problem extends to the framework of the Digital Personal Data Protection Act 2023, which governs the processing of personal data based on consent and lawful purpose. Even though the Act ensures safeguards for the “Data Principal,” it does offer mandates for processing based on consent. Nevertheless, the scope of the Act for dealing with deepfakes is restricted by certain exceptions, such as when the data is shared publicly by the person themselves or through a legal duty. In this regard, deepfakes are created by making use of publicly available images.

While addressing the present laws and framework that cover the ambit of deepfake content, it is important to observe the existing judicial perspective which closely discusses the technology based harassment of women. When[1] [2] [3]  dealing with the angle of harassment laws for women, it is governed under the Sexual Harassment of Women at Workplace Act 2013^[xix] which offers a framework for redressal through Internal and Local Complaints Committees. Sexual harassment has also been recognised as a violation of the fundamental right guaranteed under Articles 14, 15, and 21 of the Constitution. Unwelcome conduct of a sexual nature, such as showing pornography or creating a hostile work environment, has also been covered under the Act. It is limited to the “workplace.” In the case of Dr. Amit Kumar v. University of Delhi^[xx] rules that POSH Act shall cover harassment taking place in digital platforms like WhatsApp or Facebook. Although this case touches upon the harassment that occurs on digital platforms, it is understood that this particular provision is constricted to workspace as stated and hence, it still does not attend the issue of deepfakes in entirety. However, the drafted bill known as “Deepfake Prevention and Criminalisation Bill, 2023” aims at filling this gap by criminalizing the act of producing and distributing any deepfake materials without the consent of the individual or when the content contains sexual connotations. The concept of watermarks and national task force has been mentioned within the bill; however, since it is just a proposed bill, it only highlights the gaps that exist.

In the United States, deepfakes are also dealt with by the legal framework in a fragmented manner since it is a new and rapid introduction in the society. The US approach is innovation-responsive in nature. It combines the right of publicity under tort law, with the  targeted criminalisation and platform accountability measures. The tort law in the US is treated under the concept of the right of publicity. The right of publicity becomes important in this context as it becomes an obvious shield for safeguarding the individuals, especially the women and public figures when they are met with any non-consented and unauthorized dissemination of content harmful to their reputation and image in public. This principle reflects itself in the recent legislative actions in the U.S.. For instance, with regards to the state of Tennessee, the ELVIS Act of 2024^[xxi],the case had increased the purview of interpretation of this principle of publicity rights by including voice and likeness as a factor to determine the infringement of this right. This was a clear answer to the concerns surrounding voice cloning and synthetic performances.

Moreover, California’s actions on enacting Assembly Bill 1836^[xxii] and Assembly Bill 2602 of 2024 reflects initiative on the legislative policy ground as it restricts the use of AI for digital representation of individuals, particularly deceased performers, without their consent, as well as contractual exploitation of this technology. In terms of creating criminal liability, US is advancing towards introducing NO FAKES Act^[xxiii], which aims to establish a right of publicity at the federal level, allowing people to sue for unauthorized digital doubles such as deepfakes. ³

From the perspective of platform[4] [5] [6] [7]  liability, the current state of the law in the U.S. continues to be influenced by Section 230 of the Communications Decency Act^[xxiv],which affords broad immunity from third-party content. There are discussions about whether a failure to adopt measures to prevent the creation of harmful deepfakes could generate negligence or product liability claims.

In congruency to the platform liability, India presents Section 79 of the Information Technology Act of 2000 which provides a conditional safe harbour to the intermediaries from the liability. The condition of safe harbour only comes into play in terms of third-party content, when the due diligence has taken place and the intermediaries possess no actual knowledge of the unlawful output. It is to note that the intermediaries continue to stay protected under this provision until and unless a court order or a formal notification recognises otherwise, and thus these platforms become insulated from any liability that arise from user-generated deepfake contents and a private complaint is raised. With the recent introduction of 2026 Amendment Rules, the recognition of deepfake content and the liability attached to it has been addressed intrinsically. The liability now includes deepfake content for ascertaining liability, by covering it under the ambit of a general synthetically generated information. Therefore, a shift has been noted in interpreting the intermediary liability since the earlier IT Act structure and the recent amendment, reflecting steps taken towards protecting the users from deepfake contents and further provisioning a redressal system to an extent as well.

Limitations on a Legal and Technical Level

Tools like watermarking, tagging, keeping metadata and blockchain-based tracing have been proposed and implemented with an intent to increase the preventive safeguards on content dissemination against misuse of AI. Yet, the effectiveness of such methods is still not reflected. It is important to note that the technical solutions are still prone to removal, manipulation, or degradation through methods like re-encoding, cropping, or adversarial AI techniques. Moreso, the lack of uniform standardised global protocols limits their applicability across borders.

Interpreting the laws present both in India and the US, it is understood that in both jurisdictions, the impediment appears with regards to the legal evidentiary standpoint as the provisions don’t accommodate the deepfake content with creators being anonymous. For the authentication and the verification process being inculcated in the system, it required to be more stringent.^[xxv] Courts in the US state the necessity of independently verifiable features and methodologically sound evidence in case of contents created by tools relying on AI algorithms.^[xxvi] Furthermore, along with the obvious definition gap that is recurring in interpretation of statutes, more victim-centric remedial mechanisms need to be introduced. In the case of India, while proposals such as the Deepfake Prevention and Criminalisation Bill, 2023^[xxvii] aim to implement watermarking and consent requirements, they are still at a nascent stage and still has a long way to go to incorporate all the necessary angles for protecting women from deepfakes.

Effectiveness of Possible Solution

With regards to a sui generis legal framework to counter the rampant deepfake circulation and synthetic media-based gender violence, author Felipe Romero-Moreno proposes certain feasible regimes that could be possible solutions.  He states how the existing laws fail to have uniformity in its application, which then percolates to an ineffective mechanism for platform accountability, and proper biometric governance of sensitive data. Though the gravity is acknowledged by GDPR, a full-fledged way to counter it is not presented, which can be implemented holistically. A targeted legislation can be an effective method of addressing especially the gendered nature of non-consensual imagery. Further, a privacy by its design itself should be followed to protect the users. For greater effectiveness of the watermarks throughout different jurisdictions, an interoperable traceability should be introduced. Finally, the quintessential element of  global harmonization over the application of the laws and provisions for protecting the victims, with balanced liability regimes are proposed in this paper.^[xxviii] 

Conclusion

With the rampant misuse of artificial intelligence in creating deepfakes, the legal frameworks across the world, particularly in the U.S. and India as discussed above, face a challenge in interpreting such contents into the definitions of the existing provisions and to bring forth changes in the provisions to ascertain liability. It is to note that the majorly affected sect as of presently, stands to be women, with escalations in reports coming up with citizens, minors and public figures as observed in cases related to Rashmika Mandana and Taylor Swift. Throughout the study, comparable laws from both the nations are put into focus to see its effectiveness in addressing the regulation of such content. The observation from studying frameworks of both the nations is that the deepfakes are treated as ancillary to the existing laws rather than it being treated as a separate subject itself. In India, the action against deepfake arises across multiple legal frameworks in bits and pieces, from Section 356 of the BNS to Section 79 of the IT Act. The U.S. mirrors the pattern of a scattered recognition of deepfake contents through interpreting the state specific statutes like ELVIS Act that focus on protecting the individual identity and not to specifically address the sexualised contents created through deepfakes.

In terms of liability of the platforms where such contents circulate, Sec 79 of IT Act and Section 230 of the Communications Decency Act present the protection to the intermediaries until actual knowledge on the content is proven. The 2026 Amendment in IT act reflects steps in actively including the deepfake content regulations, which comes closest to an action being taken towards the contents. However, it also acknowledges that a victim-centric remedy is yet not observed by either nation’s frameworks. While active work on the Deepfake Bill of India and  the NO FAKES Act of the U.S. clearly shows that efforts are being made, there is still a gap that needs to be filled in terms of protection of women from such deepfake contents.

^[i]Sonam Singh and Amol Dhumane, ‘Unmasking Digital Deceptions: An Integrative Review of Deepfake Detection, Multimedia Forensics, and Cybersecurity Challenges’ (2025) 15 MethodsX 103632 https://doi.org/10.1016/j.mex.2025.103632 accessed 28 March 2026

^[ii]‘When justice fails: Why women can’t get protection from AI deepfake abuse’ United Nations News (21 March 2026) https://news.un.org/en/story/2026/03/1167174 accessed 28 March 2026

^[iii]Aishwarya Rai Bachchan v Aishwaryaworld.com & Ors, CS(COMM) 956/2025 (Delhi High Court, 9 September 2025)

^[iv]Laura Snapes, ‘AI song featuring fake Drake and Weeknd vocals pulled from streaming services’ The Guardian (18 April 2023) https://www.theguardian.com/music/2023/apr/18/ai-song-featuring-fake-drake-and-weeknd-vocals-pulled-from-streaming-services accessed 28 March 2026

^[v]Ramcharan Ramanaharan, Deepani B Guruge and Johnson I Agbinya, ‘DeepFake Video Detection: Insights into Model Generalisation-A Systematic Review’ (2025) 9(4) Data and Information Management 100099 https://doi.org/10.1016/j.dim.2025.100099 accessed 28 March 2026

^[vi]‘From 93 women victims to 900: Deepfake abuse enters alarming phase, says report’ The Economic Times (India, 2026) https://economictimes.indiatimes.com/news/india/from-93-women-victims-to-900-growth-deepfake-abuse-enters-alarming-phase-says-report/articleshow/129392158.cms accessed 28 March 2026

^[vii]‘Rashmika Mandanna deepfake video: Delhi Police registers case’ India Today (10 November 2023) https://www.indiatoday.in/india/story/rashmika-mandanna-deepfake-video-delhi-police-case-registered-2461547-2023-11-10 accessed 28 March 2026

^[viii]Aaradhya Bachchan v YouTube Channels (Delhi High Court 2023); see also ‘Fake videos still online, Aaradhya Bachchan approaches Delhi High Court again’ NDTV (3 February 2025) https://www.ndtv.com/india-news/after-fake-content-not-removed-aaradhya-bachchan-goes-to-delhi-high-court-again-7624084 accessed 28 March 2026

^[ix]Kamya Buch v JIX5A (Delhi High Court 2025)

^[x]Constitution of India 1950, art 21

^[xi]‘When justice fails: Why women can’t get protection from AI deepfake abuse’ United Nations News (21 March 2026) https://news.un.org/en/story/2026/03/1167174 accessed 28 March 2026

^[xii]‘Taylor Swift, Deepfakes, and the First Amendment: Changing the Legal Landscape for Victims of Non-Consensual Artificial Pornography’ Georgetown Journal of Gender and the Law https://www.law.georgetown.edu/gender-journal/online/volume-xxv-online/taylor-swift-deepfakes-and-the-first-amendment-changing-the-legal-landscape-for-victims-of-non-consensual-artificial-pornography/ accessed 28 March 2026

^[xiii]Thaler v Perlmutter, No 23-5233 (DC Cir, 18 March 2025)

^[xiv]Doe v xAI Corp (US District Court, 2026) (Grok/xAI class action lawsuit)

^[xv]Ramcharan Ramanaharan, Deepani B Guruge and Johnson I Agbinya, ‘DeepFake Video Detection: Insights into Model Generalisation-A Systematic Review’ (2025) 9(4) Data and Information Management 100099

^[xvi]Sonam Singh and Amol Dhumane, ‘Unmasking Digital Deceptions: An Integrative Review of Deepfake Detection, Multimedia Forensics, and Cybersecurity Challenges’ (2025) 15 MethodsX 103632.

^[xvii]Bharatiya Nyaya Sanhita 2023, s 356

^[xviii]Information Technology Act 2000, s 66E

^[xix]Sexual Harassment of Women at Workplace (Prevention, Prohibition and Redressal) Act 2013

^[xx]Dr Amit Kumar v University of Delhi WP(C) 586/2021 & CM APPL 42639/2024 (Delhi HC) (July 17, 2025)

^[xxi]Tennessee Ensuring Likeness, Voice, and Image Security (ELVIS) Act 2024

^[xxii]California Assembly Bill 1836 (2024); California Assembly Bill 2602 (2024)

^[xxiii]NO FAKES Act (proposed, US Congress)

^[xxiv]Communications Decency Act 1996, 47 USC § 230

^[xxv]Felipe Romero-Moreno, ‘Deepfake Detection in Generative AI: A Legal Framework Proposal to Protect Human Rights’ (2025) 58 Computer Law & Security Review 106162 https://doi.org/10.1016/j.clsr.2025.106162 accessed 28 March 2026.

^[xxvi]United States v Reffitt (DDC 2022)

^[xxvii]Deepfake Prevention and Criminalisation Bill 2023

^[xxviii]Supra 25