ANI v. OpenAI

The claims

Copyright infringement under India’s Copyright Act, 1957 over the storage and use of ANI’s wire copy to train ChatGPT and over its outputs — including alleged fabricated stories and interviews attributed to ANI. ANI seeks about ₹2 crore in damages and an injunction against further use of its content.

What has happened

India’s largest news agency sued OpenAI in the Delhi High Court, arguing infringement happens at the training stage: scraping, tokenising and storing protected articles involves copying, whatever the model later outputs. It also alleges ChatGPT reproduces its reporting and hallucinates stories under the ANI name. OpenAI counters that models learn statistical relationships rather than storing expression, that any copies are transient and technical, that it blocklisted ANI’s domains from crawling in October 2024, and that Delhi lacks jurisdiction because training and servers sit abroad. The court appointed independent amici curiae, framed the core questions — training-stage copying, output infringement, fair dealing under Section 52, and jurisdiction — and heard the interim injunction application across 32 hearings between November 2024 and March 2026 before reserving orders.

Key developments

• 2024-11-19 — First hearing before Justice Amit Bansal: summons issued, no immediate injunction, and the court moves to appoint amici curiae for the novel questions.
• 2025-01 — Indian publishing and digital news industry groups, including the Federation of Indian Publishers and the Digital News Publishers Association, seek to intervene in support of ANI.
• 2025-02-21 — An amicus curiae tells the court the Delhi High Court has jurisdiction to hear the case notwithstanding OpenAI’s foreign servers.
• 2026-03-27 — Arguments on the interim injunction conclude after 32 hearings spanning 16 months.
• 2026-04-02 — Justice Bansal reserves orders on ANI’s interim injunction application; the ruling is awaited.

Why it matters for training data

India is a huge English-language content market with no US-style fair use: the case tests whether LLM training fits the closed list of fair dealing exceptions in Section 52. OpenAI’s post-filing blocklisting of ANI’s domains did not end the dispute — opt-out is not a settlement. If Delhi takes jurisdiction, foreign model companies face Indian courts wherever their outputs reach Indian users, and the coming ruling will shape licensing terms across the market.

Sources

• The Leaflet: ANI v. OpenAI — everything so far
• Business Standard: Delhi HC reserves order (April 2026)
• Business Standard: amicus says Delhi HC has jurisdiction
• Prime Legal: 2026 case update

Deeper analysis

• Training-data copyright lawsuits: the landscape
• Is scraped data fair use for AI training?

← All case files Weekly lawsuit news