The Synthetic Reality: Assessing India’s Regulatory Architecture for Deepfakes

Author: Prisha Dhimmer, Bharti Vidyapeeth, New Law College, Pune

Introduction

The rising expansion of extremely realistic Deepfake cases in India have surged by 550 per cent since 2019, with losses projected to reach Rs 70,000 crore in 2024. Deepfakes are created using advanced AI, the Generative Adversarial Network (GANs). The issue of deepfakes remains a threat to India, which can influence a vast number of people on a larger scale, damaging their trust. The misuse of technology, fraud, and national security issues have turned this into a global problem requiring regulation. This blog will address this problem, explain how this technology impacts victims of deepfake, the ethical lines drawn around algorithms, the laws to combat deepfake technology in India, and challenges associated with it. The question is no longer if we should regulate AI, but how.

What is a deepfake?

Deepfakes are the artificially generated content created by advanced algorithms using artificial intelligence. First, a large data of the individual’s photo, video and audio is gathered, then the trained models refine the data. Deepfake technology offers considerable advantages by increasing global accessibility through authentic film dubbing and changing education and art by reintroducing historical figures. On the other hand, it is also used to distribute false news that can harm someone’s reputation and to mislead public opinion. It is very difficult to verify authenticity and it reduces trust in the media, therefore hampering the balance of the society.

How does it affect society?

Sumsub, a global identity verification provider, reported a 280% year-on-year rise in deepfake incidents in Q1 2024, especially in the run-up to national elections.Deepfakes are causing a disparity in the democratic harm of the society, it increases the effects of misinformation, it weakens the trust in the digital media, and it facilitates the evil of propaganda. Apart from this, Non-consensual sexual deepfakes are mostly targeting women in a malign manner, thus requiring rapid action on the part of the IT Rules, 2021. Illicit scraping of facial information is fueling the rise of deepfakes as well as cyber-bullying. This invasion of privacy results in extreme psycho-trauma. As per the National Cybercrime Reporting Portal, the complaint of women in the case of deepfake attacks is constantly rising with more than 90% of the global situations of deepfake being related to pornography. Deepfakes have a large significance in the matters of economic and social insecurity. This impersonation related to politicians, celebrities, and officials, among others, by the use of deepfakes triggers the evil of propaganda and results in the imbalanced conditions of diplomatic relations. Deepfakes related to frauds and defamation, among others, affect the economic security of the society in a manner that threatens e-governance. This resulting information or the code of impersonation results in the rise of communal tensions which leads to incitement of violence by damaging the public trust.

The Legislative Landscape

The Indian regulatory structure for AI and synthetic media had to adapt very quickly to address the challenging threat posed to society by deepfakes.

1. IT (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2025 These rules introduced a legal definition for Synthetically Generated Information (SGI). According to this, all content altered by AI must be marked with a watermark.

2. Bharatiya Nyaya Sanhita (BNS), 2023- BNS is substituting the IPC and comes with particular mechanisms for prosecution. Section 319: Cheating by Impersonation is being widely used in voice cloning in AI, and Section 353 is directed towards spreading wrong information that could cause disturbance in public peace.

3. Digital Personal Data Protection (DPDP) Act, 2023: The DPDP Act terms Deepfakes made without consent under the processing of "Personal Data" as unlawful. The penalty is very heavy at ₹250 crore.

MeitY Directives & Enforcement

The Ministry of Electronics and Information Technology, or MeitY, has actually put the onus on platforms and content creators through various important mandates:

"Disclosure Standards": The AI media should be accompanied by a disclosure statement that covers at least 10% of the display area.

Originator Metadata: Must be retained by the platform to make it possible to track back the illicit data to the originator, removing digital anonymity.

Safe Harbour Compliance: Reported deepfakes in the intermediary’s ecosystem have to be removed within 36 hours. In times of the General Elections, the removal timeframe can be only 3 hours for content reported to the Election Commission.

Judicial Activism and the Concept of "Fact-Check"

India does not have a specific "Deepfake Act," but the Judiciary has bridged this gap by establishing Personality Rights. Landmark cases like that of Amitabh Bachchan have led to the issuance of “Omnibus Injunctions” against unauthorized use of a person's image or voice in the Delhi High Court.

To further enhance the efficacy of the process, the government also established an AI-based detection system in 2025. Partnerships involving C-DAC(Centre for Development of Advanced Computing) and social media giants leverage deep learning algorithms that allow them to identify irregularities, such as the blink rate of one's eyes and the heart rate, right at the time of image

Critical Gaps and Future Needs in the framework

Despite these gains, four major challenges have remained:

1. The Context Problem: Legislation that is incapable of distinguishing between deliberate wrongdoing and legitimate uses such as satire and parody may end up censoring the Internet in undesired ways.

2. Jurisdictional Deadlocks: Under Indian laws, there is limited scope to force foreign companies to provide data in situations involving deepfakes from abroad.

3. Feasibility: Regulation is easily bypassed by open-source tools like DeepFaceLab. Currently, detection technology lags behind generation capabilities.

4. Mens Rea (Criminal Intent): It would be hard to prove that a user is guilty if they were not aware that they were sharing AI-generated images

However, the internet infrastructure can be made stronger by promoting resilient watermarks, able to withstand the process of re-encoding and also featuring “Fair Use” exemptions for artistic purposes. Internationally, for AI-assisted cybercrimes, there is a crucial need for a bilateral treaty from India, accompanied by a digital literacy mission in the nation.

Conclusion

India’s regulatory journey is at a crossroads. We have successfully built the statutory skeleton, but the forensic and technical muscle is still developing. While the law is "prepared" on paper, its real-world effectiveness is still struggling. India’s 2025 regulatory framework, spanning the BNS, DPDP Act, and IT Rules marks a pivotal shift from passive observation to active enforcement. By mandating 10% visual disclosures and traceable metadata, the law attempts to anchor a shifting digital reality. Yet, the true test lies in bridging the "technical gap" between hyper-realistic generation and lagging detection. As we codify these synthetic boundaries, we must ask ourselves, In a world where every image and voice can be perfectly mimicked, what will remain of our shared reality, and are we prepared for the day when we can no longer trust our own eyes?