The 2026 Tech Law Collision: Supreme Court Greenlights DPDP as Delhi HC Gives AI Training a Copyright Pass

The Grace Period is Over: DPDP is Live and Biting

For the past three years, Indian tech lawyers have treated the Digital Personal Data Protection (DPDP) Act, 2023 as a looming, but abstract, threat. That era ended in February 2026. The Supreme Court of India, while agreeing to hear constitutional challenges against the DPDP Act and its corresponding 2025 Rules, pointedly refused to stay the operation of the law. Coupled with the stringent privacy regulations operationalized in late 2025, the message from the apex court is unambiguous: compliance is no longer a future hypothetical. It is an immediate mandate.

But the real story of 2026 isn't just about privacy compliance—it is about the chaotic, unprecedented collision between the DPDP Act, the Copyright Act, 1957, and the newly amended intermediary rules governing Artificial Intelligence. We are watching the birth of a fragmented jurisprudence where data is simultaneously a protected personal asset and a free-for-all raw material for AI.

The RTI Muzzle: Shielding the State Under the Guise of Privacy

The core of the Supreme Court challenge lies in the DPDP Act's insidious amendment to Section 8(1)(j) of the Right to Information (RTI) Act, 2005. Previously, personal information could be disclosed under the RTI Act if a Public Information Officer deemed that the larger public interest justified the disclosure. The DPDP Act obliterates this public interest test, replacing it with a blanket exemption for any "personal data."

"By severing the public interest override in the RTI Act, the DPDP framework has effectively weaponized data privacy to shield the State from transparency. For writ practitioners, this severely limits the ability to gather evidence against state instrumentalities."

While the Supreme Court has issued notice to the Union Government, the refusal to grant a stay means this blanket RTI exemption is currently the law of the land. Practitioners advising public authorities or handling writ petitions must immediately recalibrate their evidence-gathering strategies, as state departments are already using the DPDP Act as a bulletproof vest against RTI inquiries.

The AI Paradox: Copyright Leniency Meets Privacy Rigidity

While the State tightens its grip on personal data, the courts are taking a surprisingly liberal approach to intellectual property in the age of AI. In August 2026, the Delhi High Court sent shockwaves through the IP bar with its prima facie ruling in the ANI vs. OpenAI dispute. The Court observed that OpenAI's scraping and storage of ANI’s literary works for Large Language Model (LLM) training fell within the "fair dealing" exception under Section 52(1)(a) of the Copyright Act, and therefore did not amount to infringement.

This creates a massive regulatory arbitrage that tech lawyers must intimately understand. If you are advising an AI company in India today, the Delhi High Court has effectively signaled that copyright law might not be the primary hurdle for training data. But here is the catch: Privacy law absolutely is.

If an AI developer scrapes the Indian internet for training data, that dataset inevitably contains personal data. Under the newly enforced DPDP Rules, consent for processing personal data must be free, specific, informed, unconditional, and unambiguous. There is no "fair dealing" exception for personal data under the DPDP Act analogous to Section 52(1)(a) of the Copyright Act.

The strategic takeaway for litigators is clear: If you are representing a publisher or an individual whose data has been scraped by an AI company, do not solely rely on a copyright infringement suit. Pivot your strategy to the DPDP Act. Claim a violation of the Data Principal's right to consent and demand erasure.

Intermediary Liability and the "Machine Unlearning" Nightmare

Adding fuel to the fire is the February 2026 amendment to the Information Technology (Intermediary Guidelines) Rules. The government has drastically lowered the takedown threshold for AI-generated or synthetic content and compressed the compliance timelines.

When you combine these new IT Rules with the DPDP Act's Right to Erasure, you hit the ultimate technological and legal wall: Machine Unlearning. If a Data Principal withdraws their consent and demands erasure under the DPDP Act, or if a takedown notice is issued for synthetic content under the IT Rules, how does an AI company "un-train" its neural network?

Technologically, extracting specific data from a pre-trained LLM is nearly impossible without retraining the entire model—a cost-prohibitive endeavor. Legally, the Data Protection Board (DPB) will not care about this technical limitation. The law mandates erasure unless retention is legally required.

What This Means for Practice

For corporate and technology lawyers, the developments of 2026 demand a complete overhaul of client advisories:

• Revamp Consent Architectures: Broad, bundled consent in Terms of Service is legally dead. The DPDP Rules require granular, specific consent. If your client trains AI, they need a segregated consent flow specifically for data ingestion.
• Due Diligence on Training Data: M&A and IP lawyers conducting due diligence on tech startups must now audit the provenance of AI training data. A clean bill of health on copyright is insufficient; if the data violates the DPDP Act, the entire model is a toxic asset.
• Prepare for DPB Litigation: With the Supreme Court allowing the DPDP Act to operate fully, the Data Protection Board will soon become one of the busiest tribunals in the country. Prepare for a surge in consumer-driven erasure demands weaponized against tech platforms.

India’s technology law landscape has historically been reactive. In 2026, with the simultaneous enforcement of the DPDP Act, aggressive AI intermediary rules, and precedent-setting IP judgments, the law is finally forcing the technology to adapt to it. Practitioners who still view privacy, copyright, and IT rules in silos will find themselves dangerously obsolete.