The AI Copyright Bombshell & The DPDP Reality Check: Why Delhi HC’s ‘Fair Dealing’ View Changes Everything

The Twin Pillars of Tech Law Just Shifted

If you are an Indian tech, media, or privacy lawyer, the last two weeks of judicial developments have fundamentally rewritten your immediate strategy. We are seeing a seismic clash between legacy statutes and frontier technologies, and frankly, the courts are dropping bombshells. Two major developments warrant immediate attention: the Delhi High Court’s highly controversial prima facie reading of "fair dealing" in the context of Artificial Intelligence, and the Supreme Court’s definitive refusal to stay the Digital Personal Data Protection (DPDP) Act, 2023.

Commercial AI Training as "Personal Research"? A Dangerous Stretch

Let’s start with the intellectual property earthquake. In the ongoing OpenAI v. ANI litigation, the Delhi High Court has expressed a prima facie view that training Large Language Models (LLMs) on ANI’s copyrighted literary works could potentially fall within the safe harbor of Section 52(1)(a) of the Copyright Act, 1957.

For the uninitiated, Section 52(1)(a) exempts "fair dealing" with a work for the purposes of "private or personal use, including research."

This judicial stance is, to put it mildly, legally audacious. Equating the systematic, automated scraping of exabytes of proprietary data by a multi-billion-dollar commercial entity with "private or personal use" or statutory "research" stretches the legislative intent of the 1957 Act to its absolute breaking point.

"If commercial data mining for algorithmic training is rubber-stamped as 'research' under Section 52, India effectively becomes a data-piracy haven for global AI developers, leaving domestic content creators without a remedy under Section 51."

Practice Impact for IP Litigators: This changes the entire playbook for copyright infringement suits against generative AI. You can no longer rely merely on proving access and substantial similarity. IP practitioners representing publishers and media houses must now rigorously brief courts on the commercial realities of LLM training. You must draw a hard line between non-commercial academic research and the ingestion of data to create a competing commercial product. Expect heavy reliance on international precedents (like the New York Times v. OpenAI docket in the US) to argue that commercial AI training fails the traditional four-factor fair dealing test.

The DPDP Act is Here: Stop Waiting for a Judicial Rescue

Over at the Supreme Court, a massive constitutional challenge to the DPDP Act, 2023, and the recently notified DPDP Rules, 2025, has been referred to a larger bench. But the headline for corporate lawyers isn't the referral—it is the Court’s explicit statement: "No question of stay."

Petitions by The Reporters’ Collective and the Mazdoor Kisan Shakti Sangathan rightly target the most insidious provision of the new privacy regime: Section 44(3) of the DPDP Act. This section effectively castrates the Right to Information (RTI) Act, 2005, by amending Section 8(1)(j) to create a blanket exemption for the disclosure of any personal information, stripping away the previous "public interest" override.

While the larger bench will eventually decide the vires of this amendment, the refusal to grant interim relief means the law is operational now. Masking and deletion of currently available public data will commence.

Practice Impact for Privacy Counsels: The compliance clock is officially ticking. With substantive obligations expected to bite by 13 May 2027 and Consent Manager registrations opening on 13 November 2026, corporate clients can no longer use the pending Supreme Court litigation as an excuse to delay compliance budgets. You need to start data-mapping exercises immediately.

Setting the Bar for "Meaningful Consent"

If your clients are wondering what DPDP compliance actually looks like in practice, the Supreme Court just gave us a preview in the education-tech sector. Ruling on the APAAR Scheme, the Court mandated that personal data collection must strictly comply with the DPDP Act, elevating the standard for consent under Section 6.

The Court held that consent must be "meaningful and informed," explicitly requiring an opt-out mechanism for parents or guardians. This is a direct warning to companies relying on pre-ticked boxes, bundled terms of service, or "click-wrap" apathy.

Furthermore, legal commentary is already flagging unresolved friction points, such as Section 8(5) of the DPDP Act regarding employee use of AI. If an employee inputs personal data into an enterprise AI tool, does that trigger a data breach notification under the 2025 Rules? Given that the Data Protection Board of India (DPBI) is currently processing its chairperson and member appointments, early regulatory guidance will likely be sparse, leaving lawyers to interpret these grey areas conservatively.

The Bottom Line

We are entering a volatile era of Indian technology law. On one hand, courts are bending over backward to accommodate AI innovation, even if it means torturing the definition of "fair dealing" in the Copyright Act. On the other, the Supreme Court is enforcing strict, meaningful consent standards for data privacy, while allowing a highly controversial RTI-diluting privacy law to take effect without a stay.

For practicing lawyers, the mandate is clear: prepare for aggressive IP litigation on AI data-scraping, and force your corporate clients to finalize their DPDP compliance frameworks today. The grace period is over.