India is weighing whether to create a dedicated law for AI-driven offences as Parliament increases scrutiny of social media platforms and the Centre tells the Supreme Court it is considering legislation for AI-related crimes. Legal experts interviewed by CNBC-TV18 split over whether existing laws can handle deepfakes, voice cloning, AI-generated misinformation and AI-powered scams, or whether generative AI requires a new framework.

India’s current legal framework already punishes fraud, impersonation, identity theft, extortion, forgery, cheating and criminal intimidation, including many offences now carried out with AI tools. The Information Technology Act and rules under it also place obligations on digital intermediaries, while other criminal laws apply to online offences. Former IT Secretary and former NASSCOM President R. Chandrasekhar said policymakers should first ask whether AI is only a new method for existing crimes or creates risks current law cannot address.

Chandrasekhar said digital arrest scams show the distinction. Criminals impersonate law enforcement officials to pressure victims into transferring money, and AI can make the deception more convincing through voice cloning or synthetic videos. He said the underlying offences, including impersonation, extortion and wrongful confinement, already exist in law, so policymakers must identify whether the gaps are in legislation, enforcement or technology for tracking offences, identifying offenders and tracing them.

Senior Advocate Sajan Poovayya argued that India needs legislation specifically designed for AI and generative AI because digital laws were drafted more than two decades ago. He said India has “a little too much legislation” in the area but lacks rules that deal with newer technologies. Generative AI systems can quickly create realistic text, images, audio and video at large scale, while recommendation algorithms can amplify synthetic content to millions of users within minutes.

A dedicated AI law could require disclosure when content is AI-generated, assign liability for harmful synthetic content, impose transparency duties on AI developers and platforms, define safety standards for high-risk AI systems and set rules for model deployment across sectors. The unresolved issue is responsibility: an AI developer may build the model, a business may integrate it, a user may generate the content and a platform may distribute it to millions of people.

Watermarking AI-generated content has become one proposed safeguard. Australia has discussed watermarking proposals, and other countries are examining technical standards for identifying synthetic media. Poovayya called watermarking and clear labelling of synthetic content one of the most practical safeguards, while Chandrasekhar asked who would apply the watermark: model developers, businesses deploying AI, users generating content or platforms distributing it.

The debate has gained urgency as India’s Parliamentary Standing Committee on Communications and Information Technology examines platform accountability and digital governance. Instagram has faced criticism over student protest-related content, and Meta drew attention after Facebook temporarily blocked the Prime Minister’s selfie video. The Supreme Court has expressed concern over AI-powered misinformation and digital arrest scams, while the Centre has indicated it is considering legislation targeting AI-related offences.

Experts also said regulation alone will not solve every AI-related problem. Poovayya said digital literacy will matter as synthetic videos, cloned voices and AI-generated images become more realistic. The two approaches now before policymakers are strengthening enforcement of existing laws or creating a dedicated AI legal framework, with both sides agreeing that regulation must clarify accountability across the AI ecosystem.

India is weighing whether to create a dedicated law for AI-driven offences as Parliament increases scrutiny of social media platforms and the Centre tells the Supreme Court it is considering legislation for AI-related crimes. Legal experts interviewed by CNBC-TV18 split over whether existing laws can handle deepfakes, voice cloning, AI-generated misinformation and AI-powered scams, or whether generative AI requires a new framework.

India’s current legal framework already punishes fraud, impersonation, identity theft, extortion, forgery, cheating and criminal intimidation, including many offences now carried out with AI tools. The Information Technology Act and rules under it also place obligations on digital intermediaries, while other criminal laws apply to online offences. Former IT Secretary and former NASSCOM President R. Chandrasekhar said policymakers should first ask whether AI is only a new method for existing crimes or creates risks current law cannot address.

Chandrasekhar said digital arrest scams show the distinction. Criminals impersonate law enforcement officials to pressure victims into transferring money, and AI can make the deception more convincing through voice cloning or synthetic videos. He said the underlying offences, including impersonation, extortion and wrongful confinement, already exist in law, so policymakers must identify whether the gaps are in legislation, enforcement or technology for tracking offences, identifying offenders and tracing them.

Senior Advocate Sajan Poovayya argued that India needs legislation specifically designed for AI and generative AI because digital laws were drafted more than two decades ago. He said India has “a little too much legislation” in the area but lacks rules that deal with newer technologies. Generative AI systems can quickly create realistic text, images, audio and video at large scale, while recommendation algorithms can amplify synthetic content to millions of users within minutes.

A dedicated AI law could require disclosure when content is AI-generated, assign liability for harmful synthetic content, impose transparency duties on AI developers and platforms, define safety standards for high-risk AI systems and set rules for model deployment across sectors. The unresolved issue is responsibility: an AI developer may build the model, a business may integrate it, a user may generate the content and a platform may distribute it to millions of people.

Watermarking AI-generated content has become one proposed safeguard. Australia has discussed watermarking proposals, and other countries are examining technical standards for identifying synthetic media. Poovayya called watermarking and clear labelling of synthetic content one of the most practical safeguards, while Chandrasekhar asked who would apply the watermark: model developers, businesses deploying AI, users generating content or platforms distributing it.

The debate has gained urgency as India’s Parliamentary Standing Committee on Communications and Information Technology examines platform accountability and digital governance. Instagram has faced criticism over student protest-related content, and Meta drew attention after Facebook temporarily blocked the Prime Minister’s selfie video. The Supreme Court has expressed concern over AI-powered misinformation and digital arrest scams, while the Centre has indicated it is considering legislation targeting AI-related offences.

Experts also said regulation alone will not solve every AI-related problem. Poovayya said digital literacy will matter as synthetic videos, cloned voices and AI-generated images become more realistic. The two approaches now before policymakers are strengthening enforcement of existing laws or creating a dedicated AI legal framework, with both sides agreeing that regulation must clarify accountability across the AI ecosystem.