A judgment is coming. The Delhi High Court’s decision in *ANI v. OpenAI*, now reserved as of April 21, 2026, per the Observer Research Foundation, will answer a question Indian copyright law has never directly addressed: does training a large language model on copyrighted news content constitute infringement, or does it fall within “fair dealing” under the Copyright Act?
The answer matters beyond India’s borders. The outcome here, whatever it is, becomes part of the global map of jurisdictions where AI training data has a defined legal status.
The legal gap at the center of the case
Most jurisdictions that have addressed AI training data have done so through a Text and Data Mining exception in copyright law. The EU has one. Japan’s APPI amendment addresses the data consent side of the same question. India’s Copyright Act, per analysis from the Observer Research Foundation, has no TDM exception. That’s not an oversight, it’s a statutory gap that the legislature has not yet moved to fill.
The Delhi High Court can go one of two ways. It can interpret India’s existing “fair dealing” provisions broadly enough to cover AI training, creating judicial precedent in the absence of statutory clarity. Or it can decline to extend fair dealing to LLM training, effectively ruling that AI companies training on Indian news content without licenses are infringing.
According to legal analysis cited by the Observer Research Foundation, an adverse ruling for OpenAI could establish a licensing requirement for AI training data under Indian copyright law. That analysis reflects expert interpretation of the pending judgment’s potential scope – it is not a court finding, and the actual ruling is pending.
Why this case, this jurisdiction
ANI, the Asian News International wire service, is a significant plaintiff. It’s not a single publication asserting rights over a small corpus. It’s a news distribution organization whose content has broad reach across Indian-language media. A ruling in ANI’s favor would affect not just OpenAI’s operations in India, but any AI developer training on Indian news content at scale.
India is also a large multilingual training data market. The scale of Hindi, Bengali, Tamil, Telugu, and other Indian-language content available for training purposes makes India’s legal posture on TDM practically significant, not just as a jurisdiction, but as a data source.
Context and precedent
This case follows the trajectory of copyright challenges to AI training data legality that have been filed in multiple jurisdictions over the past two years. The Anthropic settlement in the US, covered separately in today’s package, gives the India case a reference point, even though the legal frameworks are entirely different. The US cases have been civil litigation resolved through settlement; the India case involves judicial interpretation of statutory language. Both contribute to the emerging global picture of what AI training data costs.
What to watch
The judgment itself is the next material event. When the Delhi High Court issues its ruling, it will either affirm AI training as permissible under fair dealing, restrict it, or send the matter to parliament as a legislative question. Any of those outcomes is meaningful. This brief requires a follow-up immediately upon ruling. Internal flag: `[MONITOR: ANI-v-OpenAI-ruling-pending]`.
TJS synthesis
*ANI v. OpenAI* is one of India’s first significant judicial tests of AI training data legality. The absence of a TDM exception means the court’s decision carries more weight than a statutory interpretation question would, it’s filling a gap the legislature hasn’t addressed. Compliance teams at AI companies with Indian-language training corpora should treat the ruling as a potential trigger for legal exposure review, regardless of which way it goes. The uncertainty itself is the current risk.