India is grappling with a rapidly escalating deepfake crisis, largely fueled by the increasing accessibility and sophistication of smaller, more user-friendly artificial intelligence models. These readily available AI tools are enabling a surge in synthetic media, leading to widespread fraud, misinformation, and privacy violations across the nation.

The real-world impact is significant, ranging from multi-million dollar corporate scams to personal financial losses and severe reputational damage. The ease with which these models can generate convincing fake audio and video means that individuals and organizations are increasingly vulnerable, necessitating urgent and coordinated responses from both technology developers and regulatory bodies.

Surging Deepfake Incidents and Financial Scams in India

The scale of deepfake incidents in India has reached alarming levels. Sumsub, a global identity verification provider, reported a staggering 280% year-on-year rise in deepfake incidents during the first quarter of 2024, particularly in the lead-up to national elections. A McAfee survey conducted in November 2024 revealed that 75% of Indians had encountered deepfake content in the past year, with 45% knowing someone who had been defrauded by AI-generated material.

Corporate entities have not been immune to these sophisticated attacks. In January 2024, a finance executive at British engineering firm Arup was duped into transferring nearly 25 million USD through 15 transactions from its Hong Kong office. The fraud involved AI-generated deepfakes of familiar colleagues appearing on a Skype meeting, blinking naturally and speaking with authority, before the deception was uncovered.

Closer to home, a global chipmaker in India detected a deepfake during a job interview, where an applicant used synthetic technology to mimic facial movements and tone convincingly, as noted by Naveen Sharma, co-founder of Kroop AI. Beyond corporate fraud, deepfakes have infiltrated the personal sphere; in Kerala, a 73-year-old man lost 40,000 rupees after a WhatsApp call that appeared to be from a friend pleading for urgent help from Dubai.

The menace extends to public figures and celebrities. Since 2023, the Deepfakes Analysis Unit (DAU) of the Misinformation Combat Alliance has tracked hundreds of AI-generated scams. These include fake endorsements featuring prominent personalities like Ratan Tata, N R Narayana Murthy, Rahul Gandhi, Nirmala Sitharaman, and Virat Kohli. A viral investment video falsely attributed to Ratan Tata was found to be 83.8% AI-generated, which Tata later publicly debunked on Instagram. In November 2023, a deepfake of actress Rashmika Mandanna was created by morphing influencer Zara Patel’s face, highlighting the severe privacy risks, especially for women, from AI-generated explicit content, sometimes referred to as DeepNude.

Technical Challenges in Deepfake Creation and Detection

Deepfakes encompass both synthetic content created from scratch and manipulated content that alters existing videos, both forms distorting truth by inventing or rewriting it. The proliferation of these fakes is partly due to the increasing availability of smaller, more efficient AI models that require less computational power and expertise to operate. These models, often open-source or available via accessible APIs, lower the barrier to entry for malicious actors.

Detecting these sophisticated fakes is a complex and evolving challenge. Forensic experts are now studying subtle AI flaws, or AI fingerprints, to identify synthetic media. Dr. Surbhi Mathur, head of the Centre of Excellence in Multimedia Forensics at the National Forensic Sciences University (NFSU), points out that deepfake audio often lacks normal background noise, appearing ‘too clean.’ AI-generated faces frequently miss natural light variations and photo response non-uniformity (PRNU), which is the unique fingerprint left by camera sensors. Facial micro-expressions in deepfakes often lack natural blinking patterns or the nuanced movements of a person’s face or hands near the face.

While tools claiming over 90% accuracy in deepfake detection exist, their reliance on neural network-based deep learning means there is no absolute guarantee that every manipulation will be detected, according to Sandeep Shukla, director at IIIT Hyderabad. This ongoing technological arms race underscores the need for continuous innovation in detection methods. Several detection software examples are emerging, including Intel’s FakeCatcher, Hiya AI Solutions, Deepfake-o-Meter from Media Forensics Lab, Valida, and OpenAI’s Deepfake Detector. These tools analyze various anomalies to flag synthetic content, though specific pricing and detailed technical specifications are often proprietary or vary by deployment.

India’s Regulatory Response and Government Initiatives

In response to the escalating deepfake threat, the Indian government is implementing a multi-pronged strategy. While there is no dedicated deepfake law, the Information Technology Act, 2000, and its subsequent rules provide a legal framework to address cybercrimes like identity theft, cheating by personation, and privacy violations. The IT Act and its rules apply to information generated using AI tools, defining offenses and ensuring accountability.

The Ministry of Electronics and Information Technology (MeitY) has been actively engaging with industry stakeholders and social media platforms. The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, and its amendments, place specific obligations on intermediaries to not host or publish unlawful information. These rules mandate expeditious action towards removing unlawful content, with users able to report grievances to platform Grievance Officers or appeal to the Government-established Grievance Appellate Committees.

Upcoming regulations aim to mandate deepfake detection tools for Know Your Customer (KYC) and document verification processes, especially in the banking, financial services, and insurance (BFSI) sector, which Naveen Sharma identifies as particularly vulnerable. Amendments to the IT Rules, 2021, will enforce content labeling for over 10% of audio or video content and require faster takedowns – 36 hours normally, and a critical three hours during elections. These measures signify a strong push for accountability, where a person circulating or a platform hosting deepfake content, even unknowingly, could face legal action.

The Indian Computer Emergency Response Team (CERT-In) regularly issues alerts and advisories on cyber threats, including those involving AI. An advisory on deepfake threats and protective measures was published in November 2024. Furthermore, MeitY funds several research projects aimed at enhancing detection capabilities, such as Project Saakshya (IIT Jodhpur/Madras) for real-time detection across images, video, and audio; AI Vishleshak (IIT Mandi & Himachal Forensics) for explainable AI in deepfakes and signature forgery; and a Voice Deepfake Detection System (IIT Kharagpur) focused on AI-generated voice impersonation. These initiatives are crucial as India hits a 50 billion USD AI startup funding record, accelerating the IndiaAI Mission.

The Ministry of Home Affairs (MHA) has also established the Indian Cyber Crime Coordination Centre (I4C) and launched the National Cyber Crime Reporting Portal (cybercrime.gov.in) with a toll-free helpline number ‘1930’ to enable public reporting of cybercrimes, including financial frauds. These efforts underscore a comprehensive approach to combating the deepfake menace, integrating legal, technological, and public awareness components. The rapid evolution of AI technology, however, means that laws, detection tools, and public awareness must adapt at an equally swift pace, as scammers continuously refine their methods.

Analysis: The Accessibility Factor and Future Implications

The core of India’s escalating deepfake problem lies in the democratization of AI technology. While advanced AI models like those from Google, Meta, or Anthropic receive significant attention, it is often the smaller, more accessible models and tools that empower a broader range of actors to create convincing synthetic media. These models, often open-source or available through low-cost services, do not require extensive technical expertise or expensive hardware, making deepfake generation a relatively simple task for many. This accessibility is a double-edged sword: it fuels innovation but also significantly lowers the barrier for malicious use, leading to a proliferation of deepfakes that can quickly go viral, especially on social media platforms.

The implications for trust in digital information are profound. As deepfakes become harder to distinguish from reality, public trust in news, official communications, and even personal interactions erodes. This erosion of trust can destabilize democratic processes, exacerbate social divisions, and create an environment ripe for sophisticated fraud. The government’s focus on mandating detection tools for critical sectors like BFSI and enforcing content labeling is a necessary step, but the challenge remains in keeping pace with the rapid advancements in AI generation capabilities. Furthermore, the need for training police and judges in deepfake detection and its limitations, as highlighted by Sandeep Shukla, is critical to ensure effective legal recourse and deterrence.

Frequently Asked Questions

What is a deepfake and how is it created?

A deepfake is synthetic media, typically video or audio, that has been manipulated or generated by artificial intelligence to depict someone saying or doing something they never did. It is created using deep learning algorithms, often neural networks, that can learn patterns from existing data (like a person’s face or voice) and then generate new, highly realistic content based on those patterns.

Why are smaller AI models contributing to India’s deepfake problem?

Smaller AI models, often open-source or available through user-friendly interfaces, require less computational power and technical expertise to operate. This increased accessibility lowers the barrier to entry for individuals to create deepfakes, making the technology available to a wider range of actors, including those with malicious intent, thereby accelerating the spread of synthetic content.

What measures is the Indian government taking to combat deepfakes?

The Indian government is implementing several measures, including leveraging the IT Act, 2000, and IT Rules, 2021, to hold intermediaries accountable for unlawful content. MeitY is funding research projects for deepfake detection, issuing advisories, and planning new regulations that mandate detection tools for KYC processes and enforce content labeling and faster takedowns. The MHA operates the National Cyber Crime Reporting Portal for public reporting of incidents.

Conclusion

The escalating deepfake menace in India, driven by the increasing accessibility of smaller AI models, presents a formidable challenge to national security, economic stability, and individual privacy. While government initiatives and technological advancements in detection are underway, the speed at which AI technology evolves demands a continuous, agile, and collaborative response. Effective deterrence will require not only robust legal frameworks and advanced detection tools but also widespread public awareness and education to equip citizens against increasingly sophisticated digital deceptions.