Author: P.S. Yamini a BA.LLB(HONS) Student At Bharath Institute Of Law, Chennai.
ABSTRACT
Artificial intelligence has transformed digital communication, but deepfakes present serious challenges involving privacy, identity, reputation, cybercrime and democratic integrity. Deepfakes use artificial intelligence to generate or manipulate images, videos and audio recordings that convincingly depict people doing or saying things that never occurred.
Although the technology has legitimate applications in education, entertainment and research, its misuse can facilitate fraud, harassment, non-consensual intimate imagery and misinformation. India addresses these risks through criminal legislation, information technology laws, constitutional protections and intermediary regulations. However, the absence of a comprehensive legal framework specifically addressing deepfakes creates uncertainty concerning liability, compensation, evidence and effective remedies.
This article examines India’s existing legal framework, analyses the balance between freedom of expression and individual dignity, identifies enforcement challenges and proposes reforms to strengthen victim protection without unnecessarily restricting technological innovation.
Keywords: Artificial Intelligence, Deepfakes, Cybercrime, Privacy, Digital Evidence, Constitutional Law, Indian Legal Framework.
1. INTRODUCTION
Imagine receiving a video call from a family member urgently requesting money. The face appears genuine, the voice sounds familiar, and the situation seems convincing. Later, you discover that the person never made the call. Artificial intelligence has been used to imitate their appearance and voice to commit fraud.
Deepfakes are digitally generated or manipulated images, videos and audio recordings that imitate real individuals or events. They can fabricate statements, create non-consensual intimate imagery, damage reputations and manipulate public opinion. Unlike conventional misinformation, deepfakes exploit the tendency to trust visual and auditory evidence.
The issue is both legal and constitutional. A fabricated recording may violate privacy, cause financial loss and undermine democratic decision-making. At the same time, artificial intelligence supports creativity, journalism, accessibility and education. Regulation must therefore distinguish harmful deception from legitimate expression.
India addresses deepfake-related conduct through the Information Technology Act, 2000, the Bharatiya Nyaya Sanhita, 2023 (BNS), the Bharatiya Sakshya Adhiniyam, 2023 (BSA), and the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, as amended. Nevertheless, applying existing provisions to emerging forms of synthetic media can create uncertainty. This article argues that India needs a clearer, accessible and rights-based framework for addressing deepfake harms.
2. MEANING AND IMPACT OF DEEPFAKES
The term “deepfake” combines deep learning with fabricated content. Artificial intelligence systems can analyse photographs, voice recordings and videos to generate convincing representations of individuals. Deepfakes are not inherently unlawful. Filmmakers may use them for creative purposes, educators may use synthetic media to explain complex subjects, and voice synthesis can assist people who have difficulty speaking. Problems arise when such technology is used deceptively, without consent or to violate another person’s rights.
First, deepfakes facilitate financial fraud by imitating business executives, relatives or public officials. Second, fabricated intimate imagery can cause humiliation, harassment and lasting reputational harm. Third, false political speeches and fabricated recordings may mislead voters, inflame social tensions and undermine public confidence. Deepfakes also threaten the reliability of genuine evidence. Once fabricated recordings become convincing, wrongdoers may dismiss authentic material as artificial. This is known as the “liar’s dividend”: false material may be accepted as genuine, while genuine evidence may be denied as fabricated. Effective law must address both risks without treating all AI-generated content as unlawful.
3. EXISTING LEGAL FRAMEWORK IN INDIA
3.1 Information Technology Act, 2000
The Information Technology Act, 2000, provides several provisions relevant to deepfake-related wrongdoing. Section 66C addresses identity theft involving fraudulent or dishonest use of another person’s electronic signature, password or other unique identification feature. Section 66D concerns cheating by personation through a communication device or computer resource. These provisions may apply where deepfakes facilitate impersonation and fraud, provided the statutory requirements are satisfied.
Section 66E addresses specified violations of privacy involving images of a person’s private areas. Sections 67 and 67A concern obscene and sexually explicit material transmitted or published electronically, respectively. Their application to synthetic material depends on the facts and relevant legal elements.
Section 79 provides conditional protection to intermediaries for third-party information, subject to statutory requirements and due diligence. Section 69A permits blocking access to information on specified grounds under the applicable legal procedure.
Although these provisions are valuable, they do not establish a single offence covering every harmful deepfake. Fabricated content may damage someone’s reputation even when it does not involve financial fraud or sexually explicit material. Victims may consequently face uncertainty about the most appropriate legal remedy.
3.2 Bharatiya Nyaya Sanhita, 2023
The BNS replaced the Indian Penal Code, 1860, with effect from 1 July 2024. Several provisions may apply to deepfake-related offences.
Section 318 addresses cheating, while section 319 concerns cheating by personation. Sections 336 and 340 deal with forgery and the use of forged documents or electronic records as genuine. Their application to synthetic media depends on the statutory definitions and circumstances.
Section 356 addresses defamation, section 351 concerns criminal intimidation, and section 308 deals with extortion. Other provisions may apply where the conduct involves offences against women or other forms of unlawful exploitation.
However, creating an artificial image does not automatically establish a criminal offence. Investigators must prove the relevant conduct and required legal elements. A specific, carefully drafted provision addressing harmful deepfakes could improve clarity, provided it includes precise definitions and safeguards against Misuse.
3.3 Information Technology Rules and the 2026 Amendments
The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, establish due diligence obligations for intermediaries. Amendments notified through G.S.R. 120(E), dated 10 February 2026, introduced specific requirements concerning synthetically generated Information .
The amended framework addresses synthetic audio-visual content, unlawful material, certain user declarations and verification obligations, and labelling or metadata requirements. The applicable duties depend on the relevant rule, type of content and circumstances. These measures can help users identify synthetic material and provide victims with mechanisms to seek action. However, labels may be removed when content is copied, and detection tools may fail to identify sophisticated manipulation. Content shared outside covered services also presents enforcement difficulties.
4. PRIVACY, DIGNITY AND FREEDOM OF EXPRESSION
4.1 Constitutional Protection of Privacy
Article 21 of the Constitution protects life and personal liberty. In Justice K.S. Puttaswamy (Retd.) v. Union of India, the Supreme Court recognised privacy as a fundamental right connected with dignity, autonomy and personal liberty.
Deepfakes may interfere with these interests by appropriating a person’s face, voice or identity without consent. Fabricated intimate content can be particularly damaging because victims lose control over how their identities are represented.
The Digital Personal Data Protection Act, 2023, may also be relevant where identifiable images or voice data fall within its scope. Its application depends on the statutory provisions, applicable exemptions and commencement arrangements. It does not replace all remedies available for impersonation, defamation or other unlawful conduct.
4.2 Freedom of Speech and Expression
Article 19(1)(a) guarantees freedom of speech and expression, subject to the reasonable restrictions permitted by Article 19(2). Regulation must protect individuals without unnecessarily suppressing lawful expression.
In Shreya Singhal v. Union of India, the Supreme Court struck down section 66A of the
Information Technology Act because its vague and overbroad restrictions violated freedom of expression. This decision highlights the need for precise drafting and safeguards against arbitrary content removal.
Satire, parody, artistic expression, political commentary and journalism may legitimately use synthetic techniques. A blanket prohibition on AI-generated content could discourage innovation and suppress criticism. Regulation should instead consider context, intent, likely deception and harm. A clearly labelled parody differs significantly from a fabricated confession deliberately presented as genuine. Targeted and proportionate measures are preferable to indiscriminate censorship.
5. DEEPFAKES AND THE LAW OF EVIDENCE
The BSA governs the admissibility and proof of evidence, including electronic and digital records. Sections 61–63 address electronic records and relevant requirements for proving them, including certification where applicable. Deepfake cases raise an important distinction between admissibility and authenticity. A recording may satisfy procedural requirements but still be manipulated. Conversely, genuine evidence may be wrongly dismissed as artificial. Certification alone does not conclusively establish that the events depicted actually occurred.
Courts may require forensic examination of metadata, file history, editing patterns and audio characteristics. Because detection systems can produce false positives and negatives, no automated score should replace judicial assessment of the evidence as a whole.
Investigators should preserve original files, document collection procedures, record transfers and maintain an appropriate chain of custody. Where authenticity is disputed, courts should assess expert methods, limitations and reasoning rather than relying on unexplained technical conclusions. Consistent procedures for handling synthetic evidence would strengthen fair trials and reduce the risk of wrongful findings.
6. CHALLENGES IN ENFORCEMENT
Several obstacles limit effective enforcement.
Identification of offenders: Perpetrators may use anonymous accounts, false identities or overseas services. Identifying creators and establishing intention can require technical investigation and lawful cooperation from service providers.
Rapid circulation: Harmful material can be copied and redistributed within minutes. Removing an original upload may not eliminate duplicate versions, making prompt reporting and effective remedies essential.
Technical limitations: Detection tools cannot guarantee accurate identification. Their results may vary, and increasingly sophisticated generation techniques can undermine existing methods.
Unequal access to justice: Ordinary individuals may lack the money, legal assistance or technical knowledge needed to challenge fabricated intimate imagery or reputational attacks.
Cross-border enforcement: Where creators, platforms or servers are located abroad, international cooperation may be necessary. Different legal systems and response times can delay investigations.
These difficulties show that legislation alone is insufficient. Effective enforcement requires trained investigators, technical expertise, accessible complaint mechanisms and accountable institutions.
7. RECOMMENDATIONS FOR LEGAL REFORM
First, Parliament should consider a narrowly drafted framework addressing the intentional creation or distribution of harmful synthetic representations of identifiable persons. It should define prohibited conduct, specify the required mental element, provide proportionate penalties and protect legitimate expression.
Second, victims should have accessible civil remedies, including appropriate interim injunctions, removal orders and compensation. Procedures should accommodate cases in which the offender’s identity is initially unknown, while protecting due process.
Third, platforms should provide transparent grievance mechanisms for impersonation and non-consensual intimate imagery. They should communicate decisions clearly, preserve information where legally required and offer suitable review procedures. Serious cases should receive timely attention.
Fourth, reliable standards for labelling and content provenance should be encouraged. Labels and metadata can help identify synthetic material, but neither watermarking nor automated detection is infallible. Technical safeguards must form part of a broader legal response.
Fifth, law-enforcement agencies and courts need training in digital forensics. Specialist laboratories, qualified experts and consistent evidence-preservation procedures can improve investigations and reduce reliance on unreliable conclusions.
Sixth, regulation must preserve freedom of expression. Safeguards should protect legitimate satire, parody, journalism, research, education and artistic activity, subject to applicable law and the rights of others. Restrictions should be reasoned, proportionate and open to appropriate review.
Finally, public awareness should be strengthened. Individuals should independently verify urgent requests for money, avoid forwarding unverified recordings and report suspected impersonation. Educational institutions, businesses and public authorities can support digital-literacy programmes.
8. Conclusion
Deepfakes present India with significant challenges involving privacy, criminal liability, freedom of expression and electronic evidence. Existing legislation provides remedies for several forms of fraud, impersonation, defamation and unlawful digital content. The 2026 amendments to the Information Technology Rules further strengthen the framework governing synthetically generated information.
However, practical difficulties remain in identifying offenders, establishing authenticity and securing timely relief. India must therefore combine precise legal provisions with effective enforcement, accessible remedies, reliable forensic methods and constitutional safeguards. The objective should not be to prohibit AI technology but to prevent its harmful misuse. India’s readiness will ultimately depend on whether the law can protect individual dignity and public trust while preserving lawful expression and technological innovation.
