Authored By: DEBDIP BANERJEE
SOA National Institute of Law
ABSTRACT
The rapid advancement of artificial intelligence has led to the emergence of deepfake technology, which enables the creation of hyper-realistic but fabricated audio-visual content. One of the most disturbing manifestations of this technology is the creation and circulation of AI-generated deepfake sexual content, disproportionately targeting women and minors. While the Bharatiya Nyaya Sanhita, 2023 (BNS) claims to modernise India’s criminal law framework, it does not expressly recognise or define deepfake sexual offences. Instead, such acts are sought to be addressed through traditional provisions relating to obscenity, defamation, sexual harassment, and outraging modesty, which were designed for physical-world offences.
This assignment critically examines whether the existing provisions of the BNS are adequate to deal with AI-generated deepfake sexual content. It analyses the doctrinal challenges surrounding mens rea, attribution of liability, consent, anonymity, and jurisdiction in cyberspace. The paper further evaluates the overlap and gaps between the BNS and allied digital laws, highlighting enforcement deficiencies and the risk of under-criminalisation. Through an examination of landmark judicial pronouncements and comparative legal approaches from foreign jurisdictions, the study argues that the present framework is insufficient to address the unique harm caused by deepfake sexual abuse.
The assignment ultimately advocates for the creation of a distinct statutory offence within the BNS to address AI-generated sexual exploitation. It proposes a victim-centric and technology-sensitive reform model that balances individual rights, technological innovation, and effective criminal accountability. The research underscores the urgent need for legislative clarity to ensure that criminal law remains responsive to evolving forms of digital harm.
Keywords: Deepfake Sexual Content, Artificial Intelligence and Criminal Law, Bharatiya Nyaya Sanhita, 2023, Cyber Sexual Abuse, Mens Rea, Legal Reform
- INTRODUCTION
Deepfake technology, powered by advanced machine learning and generative adversarial networks (GANs), enables the creation of hyper-realistic synthetic audio-visual content that is often indistinguishable from authentic recordings. While the technology has legitimate uses in entertainment, education, and accessibility, its weaponisation in the form of non-consensual sexual content represents one of the most alarming contemporary threats to personal autonomy and dignity. Unlike traditional sexual offences that presuppose physical proximity or bodily violation, deepfake sexual abuse is executed in digital space; yet the absence of physical contact does not diminish the gravity of harm inflicted upon victims. Courts in India have increasingly recognised that psychological trauma, reputational destruction, and social ostracization constitute serious legal injuries deserving protection under criminal law.[1]
The enactment of the BNS marked a significant legislative shift, replacing the colonial-era Indian Penal Code, 1860 with the stated objective of modernising substantive criminal law. The BNS introduces reforms in sentencing, organised crime, and gender-based offences; however, it does not expressly criminalise AI-generated deepfake sexual content. In practice, prosecutors are compelled to rely upon general provisions relating to sexual harassment, obscenity, defamation, and criminal intimidation.[2] Such reliance reveals a structural mismatch between technology-driven harms and provisions drafted for conventional offences.
Therefore, the central question in contemporary criminal jurisprudence is whether a legal framework designed for physical-world offences can effectively address synthetic sexual exploitation in cyberspace. The BNS, despite its reformist intent, remains silent on this technology-driven abuse.
- GENERAL THEORIES RELATED TO THE AREA OF RESEARCH
Criminal liability has historically been structured around the twin pillars of actus reus (the guilty act) and mens rea (the guilty mind).[3] The classical formulation presupposes a voluntary human act that produces a legally cognisable harm, accompanied by a blameworthy mental state. In the context of AI-generated deepfake sexual content, however, identifying the guilty act and guilty mind becomes complex. The “act” is not a physical intrusion but the deployment of algorithmic tools to fabricate synthetic intimacy; the “mind” may be diffused across multiple actors, including the creator of the model, the user generating the content, and the distributor who amplifies it online.[4] This fragmentation complicates attribution of culpability and challenges traditional notions of direct intention and knowledge.
The difficulty increases because AI tools are now widely accessible. Deepfake generation no longer requires advanced technical expertise, blurring the distinction between primary perpetrators and secondary participants. Classical doctrines such as common intention and abetment assume identifiable coordination among individuals.[5] Theories of extended or derivative liability therefore become central in assessing responsibility within decentralised digital networks.[6]
Closely connected to harm theory is the dignitarian model of criminal law, which treats certain offences as violations of intrinsic human worth.[7] Non-consensual deepfake pornography degrades and objectifies individuals, disproportionately targeting women and marginalised communities. Scholars argue that digital sexual abuse perpetuates structural gender inequality by transforming a person’s image into a tool of humiliation without consent.[8] .
- LEGAL CONCEPTS AND RELEVANT CASE LAWS
India’s current legal framework is not completely silent on deepfake sexual abuse but relies on general offences under the BNS such as sexual harassment, obscenity, defamation, and criminal intimidation. These provisions were designed for physical or traditional misconduct, not AI-generated content. This raises a key question: can creating synthetic sexual images of a person without consent be treated as a criminal “act”?
The offence of outraging modesty historically centres upon conduct that violates a woman’s bodily integrity or sexual dignity.[9] While digital manipulation lacks physical contact, courts have progressively interpreted modesty and dignity as extending beyond corporeal invasion. The harm inflicted by deepfake pornography lies in its involuntary sexualisation of identity. The victim’s image is digitally superimposed onto explicit content, thereby constructing a false sexual narrative capable of permanent circulation.
In such circumstances, the injury is reputational, psychological, and social—yet no less severe than physical harassment. Parallel reliance is often placed upon the Information Technology Act, 2000, particularly provisions criminalising the electronic publication or transmission of obscene material.[10]
The constitutional dimension of deepfake sexual abuse significantly strengthens the case for robust criminal intervention. Article 21 of the Constitution guarantees the right to life and personal liberty, which has been judicially expanded to encompass privacy, dignity, and informational autonomy.
In K.S. Puttaswamy v. Union of India, a nine-judge bench of the Supreme Court unequivocally affirmed that privacy is a fundamental right intrinsic to human dignity.[11] The Court recognised informational control as central to autonomy, emphasising that individuals must retain authority over the dissemination and representation of their personal data and identity. Deepfake pornography directly subverts this principle by stripping individuals of control over their digital likeness and imposing a fabricated sexual persona without consent.
Judicial sensitivity toward digital sexual exploitation is further evident in State of West Bengal v. Animesh Boxi, where the Calcutta High Court addressed the circulation of morphed explicit images of a woman.[12]
- CRITICAL ANALYSIS
The rising spread of AI-generated deepfake sexual content reveals a major gap in India’s criminal law. Although courts have shown sensitivity in cases involving morphing, cyber harassment, and non-consensual sharing of intimate images, judicial interpretation cannot replace clear legislation. The absence of a specific offence under the BNS leads to fragmented prosecution, legal uncertainty, and inconsistent sentencing. This section examines these gaps and suggests reforms based on comparative practices and constitutional principles.
Fragmented Prosecution and Doctrinal Inconsistency
At present, prosecutors rely on different provisions under the BNS and the IT Act, such as sexual harassment, obscenity, defamation, and criminal intimidation. This patchwork approach creates three key problems.
First, it misrepresents the harm. Deepfake sexual abuse mainly violates consent and personal identity, but obscenity laws focus on public morality and defamation focuses only on reputation. As a result, the law addresses the harm indirectly rather than recognising it as a direct attack on sexual dignity.
Second, using different charges for similar cases leads to inconsistent punishment. Some courts may treat such cases as obscenity while others treat them as defamation, resulting in uneven sentencing and weaker deterrence.
Third, proving the offence becomes more difficult. Because there is no specific law for deepfakes, prosecutors must stretch existing provisions to fit new technology, which allows defence lawyers to challenge cases based on gaps in the law.
The Problem of Mens Rea in AI-Generated Abuse
Deepfake offences also complicate the classical requirement of mens rea. In conventional crimes, the perpetrator both commits the act and possesses the guilty mind. In AI-assisted abuse, multiple actors may be involved like, the developer who created the generative software, the user who inputs the victim’s image, the individual who circulates the content.
Attribution of intent becomes layered. Should liability attach only to the person who knowingly generates non-consensual sexual content? What about individuals who recklessly share such content without verifying authenticity? The current framework provides no clear answers.
Comparative Analysis: United Kingdom and United States
India’s reactive framework contrasts with international trends that explicitly criminalise deepfake sexual abuse. The United Kingdom, through its Online Safety reforms, recognises synthetic intimate images as a distinct form of harm requiring specific statutory treatment. Similarly, several U.S. states, including California, Texas, and Virginia, have enacted laws criminalising non-consensual deepfake pornography, with some also providing civil remedies. The comparative lesson is clear: clear statutory recognition reduces ambiguity, strengthens deterrence, and affirms the seriousness of digital sexual exploitation—something India’s reliance on general provisions presently lacks.
Legislative Reform Proposal
To address the existing lacuna, the BNS should incorporate a distinct provision.
- Whoever, by means of artificial intelligence, digital manipulation, or synthetic media technology, creates, publishes, or transmits any sexually explicit representation of a person without that person’s consent, shall be punished.
- Where such act is committed with intent to cause humiliation, harassment, extortion, or sexual exploitation.
- Where the victim is a minor, the offence shall be cognizable and non-bailable
- Any person who knowingly distributes such content, having reason to believe that it is non-consensual synthetic sexual media, shall be punished.
- It shall be a defence if the accused proves that the content was created for bona fide educational, artistic, or lawful investigative purposes and without intent to cause harm.
Additional Procedural Safeguards
Reform must extend beyond substantive criminalisation. The following procedural mechanisms are essential: Time-bound content removal orders through fast-track cyber courts, Victim anonymity protections during trial, Mandatory digital forensic protocols for authentication of synthetic media, Platform compliance obligations for swift takedown upon notice. Such safeguards ensure that criminalisation is not symbolic but practically enforceable.
- CONCLUSION
The rise of AI-generated deepfake sexual content poses a serious challenge to modern criminal law. AI can create highly realistic explicit images or videos of individuals without their consent, harming their dignity, privacy, and reputation. Once shared online, such content spreads quickly and is difficult to remove, often causing lasting psychological and social harm.
Although the Bharatiya Nyaya Sanhita seeks to modernise India’s criminal law, it does not adequately address the specific harms caused by AI-generated sexual exploitation. Existing provisions on obscenity, defamation, voyeurism, or cyber harassment may be applied, but they were designed for traditional misconduct rather than technologically created manipulation. As a result, courts and prosecutors often rely on broad interpretations of existing offences, which can create uncertainty regarding consent, intent, and proof of responsibility.
The absence of a specific legal provision targeting AI-generated sexual content also weakens deterrence. Deepfake technology is becoming increasingly accessible and inexpensive, allowing harmful content to be produced and shared within minutes. This issue also raises important constitutional concerns. The circulation of fabricated sexual imagery directly threatens the rights to dignity and privacy protected under Article 21 of the Constitution of India. Therefore, clear legislative reform is needed to criminalise the creation and distribution of non-consensual AI-generated sexual content and to introduce victim-centred protections such as quick takedowns, identity protection, and stronger penalties.
6.Reference(S):
[1] See, e.g., State of West Bengal v. Animesh Boxi, 2018 SCC Online Cal 3182 (India).
[2] Bharatiya Nyaya Sanhita, 2023, §§ relating to sexual harassment and obscenity (India).
[3] GLANVILLE WILLIAMS, TEXTBOOK OF CRIMINAL LAW 71–72 (2d ed. 1983).
[4] See Danielle Keats Citron & Robert Chesney, Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security, 107 CALIF. L. REV. 1753, 1760–63 (2019).
[5] Bharatiya Nyaya Sanhita, 2023, § 3(5) (India) (common intention provision).
[6] See ANDREW ASHWORTH, PRINCIPLES OF CRIMINAL LAW 185–89 (7th ed. 2013).
[7] ANDREW ASHWORTH, PRINCIPLES OF CRIMINAL LAW 34–36 (7th ed. 2013).
[8] Mary Anne Franks, Sexual Harassment 2.0, 71 MD. L. REV. 655, 670–75 (2012).
[9] See Rupan Deol Bajaj v. K.P.S. Gill, (1995) 6 S.C.C. 194 (India) (interpreting outrage of modesty).
[10] Information Technology Act, 2000, No. 21, Acts of Parliament, Sec- 67, 67A (India).
[11] K.S. Puttaswamy v. Union of India, (2017) 10 S.C.C. 1, ¶¶ 297–300 (India).
[12] State of West Bengal v. Animesh Boxi, 2018 SCC Online Cal 3182 (India).
