Delhi HC Grants 'Fair Dealing' Pass to AI Training, While Supreme Court's DPDP Limbo Forces Corporate India to Comply Blindly

The Generative AI Copyright Shield: Delhi High Court Stretches Section 52(1)(a)

For IP practitioners watching the collision between generative AI and copyright law, the wait for Indian jurisprudence is over—and the initial results are highly alarming for content creators. In a landmark 2026 development, the Delhi High Court has held, prima facie, that OpenAI’s ingestion and storage of ANI’s literary works to train the Large Language Models (LLMs) underlying ChatGPT falls squarely within the "fair dealing" exception of Section 52(1)(a) of the Copyright Act, 1957.

This is a seismic shift. Section 52(1)(a) exempts fair dealing with any work for the purposes of "private or personal use, including research." By categorizing the mass scraping and ingestion of copyrighted news media for commercial LLM training as "research," the Delhi High Court has effectively handed Big Tech a blank cheque.

"The court's willingness to stretch the statutory bounds of 'research' to cover commercial algorithmic training neuters the primary leverage media houses and authors had against AI developers. We are witnessing the judicial prioritization of technological advancement over traditional IP monopolies."

What this means for your practice: If you represent media houses, publishers, or digital content creators, your traditional cease-and-desist letters citing Section 14 (exclusive right to reproduce) are now severely compromised. The burden has shifted. You can no longer rely on the mere act of data scraping as per se infringement. Instead, litigators must now focus on output-level infringement—proving that the AI's generated responses are substantially similar to the protected expression of the original works, rather than fighting the input-level ingestion. For tech lawyers representing AI startups, this ruling is your new shield against intermediary liability and copyright claims.

The DPDP Catch-22: Constitutional Scrutiny Meets Relentless Compliance

While tech companies celebrate the Delhi High Court's copyright stance, their regulatory teams are facing a nightmare courtesy of the Supreme Court. The Apex Court has referred multiple challenges against the Digital Personal Data Protection (DPDP) Act, 2023 and the DPDP Rules, 2025 to a larger bench, acknowledging that the constitutional issues require deeper scrutiny. Yet, crucially, the Court declined to stay the Act’s operation.

The core of the constitutional challenge—spearheaded by the Reporters Collective, NCPRI, and Mazdoor Kisan Shakti Sangathan—revolves around the DPDP Act’s backdoor amendment to the Right to Information (RTI) Act, 2005. Specifically, Section 44(3) of the DPDP Act amends Section 8(1)(j) of the RTI Act, effectively creating a blanket exemption for the disclosure of any personal information, obliterating the previous "public interest" test.

Because the Supreme Court refused interim relief against the masking or deletion of already-available public data, corporate counsel and Data Fiduciaries are trapped in a regulatory Catch-22. You must implement sweeping, expensive compliance frameworks for a law that might be fundamentally rewritten by a constitutional bench.

What this means for your practice: You cannot advise your clients to "wait and watch." The DPDP Rules 2025 are already in force, with phased compliance deadlines hitting hard between now and May 2027. The Consent Manager framework goes live in November 2026. Advise your corporate clients to proceed with data mapping and overhauling their notice-and-consent architecture immediately. If the Supreme Court eventually strikes down the RTI amendments, it will likely be surgical; the core obligations of Data Fiduciaries under Section 8 (general obligations) and Section 9 (children's data) will survive.

The APAAR Ruling and the Intersections of AI, Employment, and Privacy

Adding to the privacy compliance burden, the Supreme Court has tightly linked the education sector's APAAR (Automated Permanent Academic Account Registry) scheme to the DPDP Act. The Court mandated that the collection and retention of student personal data requires meaningful parental consent.

This ruling signals that the judiciary will not allow state-run or public-interest tech initiatives to bypass strict DPDP scrutiny. EdTech startups and universities must urgently audit their data pipelines. A simple "I Agree" checkbox will not survive judicial scrutiny under the new regime; consent must be verifiable, specific, and capable of being withdrawn seamlessly.

Furthermore, as we move deeper into 2026, the intersection of the DPDP Act and enterprise AI is causing friction. Section 8(5) of the DPDP Act requires Data Fiduciaries to protect personal data in their possession. With employees increasingly feeding proprietary and personal data into enterprise AI tools, companies face massive exposure. If an employee inputs a client's personal data into an LLM, the Data Fiduciary has arguably breached Section 8(5) by failing to implement reasonable security safeguards against unauthorized processing.

The Bottom Line: Indian technology law in 2026 is defined by aggressive judicial interpretation. The Delhi High Court is insulating AI development from copyright claims, while the Supreme Court is forcing strict, un-stayed compliance with a privacy regime whose constitutional validity remains sub judice. For practicing lawyers, the mandate is clear: draft aggressively protective AI use policies for employers, pivot IP litigation strategies toward output-infringement, and push corporate clients toward strict DPDP compliance despite the lingering Supreme Court challenges.