The Tech Law Paradox: Delhi HC Gives AI a 'Fair Dealing' Pass While the DPDP Act Faces Constitutional Fire

A Schizophrenic Approach to Digital Regulation

If the last few months of Indian technology and intellectual property law have proven anything, it is that we are operating in a state of regulatory schizophrenia. On one hand, the State is tightening the noose on platforms and data fiduciaries with draconian compliance timelines. On the other, the judiciary is showing a surprising, almost liberal flexibility toward generative artificial intelligence under legacy IP frameworks.

For technology lawyers, IP practitioners, and in-house counsel, the late 2025 and early 2026 developments are not just academic updates—they are immediate, practice-altering shifts. The operationalization of the Digital Personal Data Protection (DPDP) Act, controversial amendments to the IT Rules, and a landmark Delhi High Court observation on copyright and AI have fundamentally rewritten the risk matrix for digital businesses in India.

ANI v. OpenAI: A Lifeline for Generative AI

Let us begin with the most consequential IP development of 2026. In a dispute between Asian News International (ANI) and OpenAI, the Delhi High Court made a prima facie observation that is bound to send shockwaves through the media and entertainment industries. The Court noted that OpenAI’s storage and ingestion of ANI’s literary works for training its Large Language Models (LLMs) amounts to fair dealing under Section 52(1)(a) of the Copyright Act, 1957.

Why does this matter for your practice? Because the Court bypassed the temptation to demand new legislation and instead stretched a 1957 doctrine to cover 2026 technology. The rationale hinged on market substitution: because the LLM’s output does not serve as a direct market substitute for the original news reports, the ingestion for training purposes is permissible.

"This is a massive blow to copyright maximalists and a definitive victory for AI developers. It signals that Indian courts are willing to interpret 'research' and 'private study' under Section 52 expansively to protect technological innovation."

For IP lawyers advising AI startups, this is the shield you have been waiting for. However, if you represent publishers or authors, your litigation strategy must pivot immediately. You can no longer merely prove that your client's data was scraped; you must tangibly demonstrate that the AI's output is cannibalizing your client's commercial market.

The DPDP Act is Live, But Surviving on Borrowed Time?

While AI gets a judicial pass, data fiduciaries are facing the music. In November 2025, the Centre finally notified the administrative rules for the DPDP Act, operationalizing the 2023 legislation. Companies are now legally bound to enforce purpose limitation, verifiable consent mechanisms, and strict breach notifications.

But the real story is in the Supreme Court. In February and March 2026, the apex court issued notices on multiple PILs challenging the constitutionality of the DPDP Act. The core of the challenge lies in the Act's severe dilution of the Right to Information (RTI) framework—specifically, the blanket exemption introduced by amending Section 8(1)(j) of the RTI Act, which effectively shields all "personal information" from public scrutiny, regardless of public interest.

Petitioners in both the Supreme Court and the Delhi High Court are arguing that the Act violates Articles 14, 19, and 21 of the Constitution. Crucially, the Supreme Court has signaled it will dive into the jurisprudential deep end to define the boundary between public data and private data—a distinction that the DPDP Act deliberately blurs.

The Practical Takeaway: Do not let your corporate clients pause their compliance audits just because of these constitutional challenges. The Supreme Court explicitly refused to stay the operation of the Act. The law is in force. If you are an in-house counsel, your priority remains overhauling legacy data collection forms to ensure users have a clear chance to refuse consent, as mandated by the new November 2025 rules. Waiting for the Supreme Court's final verdict on the RTI conflict is a luxury your clients cannot afford.

The 3-Hour Takedown Trap

As if the DPDP compliance burden wasn't enough, 2026 also saw a lethal amendment to the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules. For cases involving "synthetically generated information" (read: deepfakes and AI-generated misinformation), the takedown window has been slashed from 72 hours to a suffocating 3 hours.

From a practical standpoint, this is a compliance nightmare that threatens the safe harbor protection granted under Section 79 of the Information Technology Act, 2000. How exactly is an intermediary supposed to receive a complaint, legally and technically verify whether content is a malicious deepfake or legitimate satire, and execute a takedown in 180 minutes?

This amendment forces the hand of tech platforms. To avoid losing their Section 79 immunity, intermediaries will have no choice but to deploy aggressive, automated algorithmic takedown tools. For technology lawyers, expect a surge in intermediary liability disputes and wrongful takedown notices. You must advise platform clients to prioritize triage mechanisms specifically for synthetic media complaints, as the margin for error is now virtually non-existent.

The Convergence of Tech Law

We are witnessing the death of the siloed tech lawyer. You can no longer practice IP law without understanding algorithmic ingestion. You cannot advise on data protection without navigating constitutional law and the RTI Act. As AI, privacy, and intermediary liability collide, the lawyers who will dominate the next decade are those who can synthesize these overlapping, and often contradictory, regulatory frameworks.