Introduction

Artificial Intelligence is no longer a distant idea. It already writes emails, approves loan applications, helps doctors read scans, and creates videos that look real but are not. [1]As AI becomes part of everyday life in India, a question keeps coming up: should the country have one dedicated law just for AI, or is it better to keep adjusting the laws we already have? This question has moved from academic debate rooms into the corridors of government, with the Ministry of Electronics and Information Technology openly discussing what an AI governance framework for India should look like. [2]This blog looks at what India is doing right now, what the real legal gaps are, and whether a separate AI law makes sense at this stage.

Background

India's engagement with AI regulation has been gradual. It started with the NITI Aayog's National Strategy for AI in 2018, [3]which focused on using AI for development goals like healthcare and agriculture, not on regulating it. As generative AI tools like ChatGPT became mainstream[4], the conversation shifted from 'how do we use AI' to 'how do we control its risks'. The government's approach has leaned towards a 'light touch' model. Instead of pausing AI development to write a brand-new law, India has stretched its existing legal tools to cover AI-related harms, while building softer, non-binding guidance alongside them. The IndiaAI Mission, [5]approved in 2024 with a budget of over ten thousand crore rupees, shows the government's bigger priority is building AI capability, with regulation not meant to slow that down.

Legal Framework

India does not have one single, standalone law that regulates AI as a technology. [6]Instead, AI is governed indirectly through a patchwork of existing laws. The Information Technology Act, 2000 remains the base law for anything digital, and was amended in February 2026 [7]through the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules. These rules created a new legal category called 'Synthetically Generated Information' or SGI, covering AI-made audio, video and images that look real. Platforms must now label such content [8]and take it down within two to three hours if unlawful, or risk losing safe harbour protection under Section 79 of the IT Act.[9]

The Digital Personal Data Protection Act, 2023 controls how personal data can be collected and used, affecting how AI models are trained. The RBI's FREE-AI framework applies to banking, credit scoring and fraud detection. The Supreme Court has released draft rules for AI inside courts, allowing it to assist with scheduling or research but not to decide a case. MeitY released the IndiaAI Governance Guidelines in early 2026, built around seven 'sutras' such as fairness and accountability, though these are not legally binding. A private member's bill, the Artificial Intelligence (Ethics and Accountability) Bill, 2025, proposes an ethics committee and mandatory bias audits, though it has not become law. The Digital India Act, [10]expected to replace the outdated IT Act, is likely to include AI-specific provisions once finalised.

Relevant Case Laws

Since India has no dedicated AI statute, courts have improvised using constitutional and common-law principles, and a clear body of case law has built up. In Justice K.S. Puttaswamy v. Union of India (2017), the Supreme Court [11]recognised privacy as a fundamental right under Article 21, shaping how AI systems handling personal data are expected to behave. In Shreya Singhal v. Union of India (2015), [12]the Court struck down Section 66A of the IT Act for being vague, a ruling that still guides how far intermediary rules, including the 2026 amendment on AI content, can go without curbing free speech.

More recently, courts have used personality rights to tackle AI misuse directly, since no deepfake-specific law exists yet. In Anil Kapoor v. Simply Life India (Delhi High Court, 2023), [13]the court protected the actor's name, voice and catchphrase 'Jhakaas' from unauthorised AI-generated use. In Jackie Shroff v. The Peppy Store & Ors. (Delhi High Court, 2024), [14]the court restrained an AI chatbot from commercially using the actor's persona, one of the first Indian orders aimed directly at an AI tool. The Bombay High Court's ruling in Arijit Singh v. Codible Ventures LLP (2024), [15]India's first judgment on generative AI misuse, restrained platforms from cloning the singer's voice without consent. These cases, along with similar orders for Amitabh Bachchan and Aishwarya Rai Bachchan, show courts actively filling the AI regulation gap case by case, using 'John Doe' orders against anonymous infringers, even while Parliament has not passed a dedicated AI law.

Analysis and Discussion

There are genuine arguments on both sides. Those in favour of a separate AI law point out that current laws were never designed with AI in mind. If an AI system denies someone a loan unfairly or a self-driving algorithm causes an accident, it is often unclear who should be held responsible: the developer, the deploying company, or the data provider. Even the personality rights cases above show the limits of relying only on courts, since each celebrity has had to file a separate suit, rather than one statutory rule applying to everyone automatically. A dedicated law could set clear rules for high-risk uses of AI, such as healthcare, hiring and credit scoring, and create one clear regulator instead of MeitY, RBI, the Supreme Court and various High Courts each shaping the rules separately.

On the other hand, there is a strong case for not rushing into a standalone law. AI technology is changing fast, and a rigid law passed today could feel outdated within a couple of years, forcing Parliament into constant amendments. India's government has repeatedly said it sees no immediate need for a separate AI law, preferring a 'techno-legal' approach combining legal rules with technical safeguards like watermarking and bias detection built into AI systems. Comparisons are often drawn with the EU's AI Act, [16]whose strict, upfront risk classifications have reportedly created heavy compliance costs that could hurt smaller Indian AI startups. India's sector-by-sector strategy, backed by a judiciary willing to grant fast interim relief, lets regulation catch up gradually with real, observed harms rather than hypothetical ones.

Recent developments suggest India may land somewhere in the middle. The February 2026 IT Rules amendment and the RBI's FREE-AI framework show India is comfortable regulating AI sector by sector as harms appear, while the Delhi and Bombay High Courts keep plugging gaps through personality rights litigation. Ongoing work on the Digital India Act suggests a broader law is still being planned, just not treated as urgent yet. Copyright is another area under review, with a DPIIT committee examining how AI firms should compensate creators, showing regulation expanding step by step rather than through one giant law.

Conclusion

India does not need to rush a single, all-covering AI law, but it cannot rely only on scattered guidelines and case-by-case court orders forever. The sensible path is a hybrid one: keep strengthening existing laws for concerns like deepfakes and data protection, let courts keep filling urgent gaps through personality rights litigation, and work steadily towards the Digital India Act to fix the larger structural gaps around liability for AI-caused harm. A standalone AI law could still help later, but only once India has enough real-world experience with AI harms to write rules that are practical rather than theoretical. For now, keeping the system flexible seems a fair trade-off between encouraging India's growing AI industry and protecting citizens from its risks.

References

2026.