Madras High Court Rules Unauthorized Use of Copyrighted Content for AI Training Raises Prima Facie Case of Infringement

The Madras High Court in India delivered a significant ruling in the case of Keshan Infotech vs Oliver Brandt, holding that the unauthorized use of proprietary content as training data for large language models or as prompts for AI-based content generation tools raises a prima facie case of copyright infringement.

The plaintiff, Keshan Infotech, operates travel website travelandtourworld.com and alleged that defendant Oliver Brandt, an Italian editor, systematically scraped its copyrighted travel content and reproduced it on Google-owned social media platforms while retaining the original website's look and feel.

Senior Advocate Rudraman Bhattacharya sought four injunctions on behalf of the plaintiff. Justice K. Kumaresh Babu noted that a prima facie reading of the plaint indicated the applicant was a travel vlogger producing original content, and that the allegations were substantiated by the respondent's social media posts annexed to the suit. The court granted all four injunctions for a period of four weeks.

The ruling reignites a critical legal debate under Indian law: does training AI models on copyrighted content without permission constitute copyright infringement, or does it fall under 'fair dealing' as defined in Section 52 of the Indian Copyright Act? Indian law currently offers no clear answer.

The decision comes amid ongoing AI copyright battles in India, including ANI's lawsuit against OpenAI. The DPIIT working paper on AI also explicitly declined to resolve whether AI training infringes copyright. The Madras HC's interim injunction signals that courts can impose AI training restraints even while the underlying legal questions remain unresolved.

This ruling adds an Asian jurisdiction perspective to a growing global divergence on AI copyright. US courts last year ruled that using purchased copyrighted works for AI training constitutes fair use, while a German court found OpenAI liable for reproducing copyrighted song lyrics through ChatGPT.