Is India's approach of relying on existing laws such as the IT Act adequate to address harms from deepfakes and synthetic media? Critically analyse.

India has no standalone AI Act. Instead it handles deepfakes and synthetic media through the IT Act, 2000, rules made under it, and non-binding guidelines. This approach works well against clearly unlawful content. It falls short against the wider loss of trust caused by lawful but misleading synthetic media.

Strengths of the existing-law approach

• Ready penal provisions: Under Sections 67A/67B, publishing or transmitting obscene material is punishable, and this includes deepfake-generated content [1].
• Quick updates through rules: The IT Rules, 2021 were amended (10 February 2026) to cover synthetically generated information (SGI). Platforms must now use technical measures against unlawful SGI and label permitted synthetic content with traceable metadata [2].
• Targeted harms: The amendment names child sexual exploitation material, non-consensual intimate imagery and impersonation [2].
• Pro-innovation: The India AI Governance Guidelines (2025) set out a risk-based, proportionate framework that builds on existing laws and avoids a heavy new statute [3].

Limitations

• Lawful harms fall outside the rules: Takedown duties apply only to illegal content. A misleading but legal deepfake breaks no law, yet it still damages democratic discourse.
• Labels are fragile: A screenshot or re-recording can strip the metadata. Shrinking attention spans also mean that people often ignore labels they can see.
• Weak legal footing: The framework rests on delegated rules and non-binding guidelines, not on a law passed after debate in Parliament [3]. This leaves the legal position uncertain.
• Free-speech risk: Broad definitions may push platforms to over-remove content, which conflicts with Article 19(1)(a).
• Reactive enforcement: Victims still depend on complaints, and harm spreads faster than takedowns.

Way forward

• A calibrated statute or dedicated chapter for high-risk uses such as elections and finance, consistent with the risk-based approach [3].
• Watermarking/provenance standards that survive editing.
• Media and information literacy and transparency/explainability, as the UNESCO Recommendation on the Ethics of AI (2021) urges [4].

On balance, existing law is a sound first line of defence against illegal deepfakes. It is not adequate against the broader decline in trust. A layered model would protect both Article 21 dignity and free expression while still supporting innovation. That model should combine targeted legislation, technical provenance and citizen literacy.