OpenAI faces Delhi court appeal over ANI news scraping for AI training

This case is one of the clearest tests yet of how a major non‑Western jurisdiction will treat news scraping for model training. The Delhi High Court is not ruling on final liability but it is willing to reopen the question of whether using ANI’s archive to train large language models can qualify as “fair dealing” under Indian copyright law. That alone signals to global AI labs that they cannot assume US‑ or UK‑style exceptions will travel unchanged to India’s courts.([medianews4u.com](https://www.medianews4u.com/delhi-hc-seeks-openai-response-on-ani-appeal-over-copyrighted-news-content/))

For the race to AGI, the stakes are less about this one news agency and more about precedent. If ANI eventually wins tougher limits or compensation, other publishers across the Global South will follow, compressing the pool of high‑quality text data that can be quietly scraped. That could force OpenAI and peers into more explicit licensing, state‑brokered data deals, or synthetic‑data strategies sooner than planned. It also sharpens the geopolitical contrast: India is both a huge AI talent base and a rights‑conscious democracy that may not simply rubber‑stamp Silicon Valley’s training norms. Labs aiming for world‑scale models will increasingly need legal strategies tailored to Delhi, not just DC and Brussels.