‘Infringement does occur’: ANI on OpenAI using its data for training

ANI challenges OpenAI in court, alleging copyright infringement over data used to train ChatGPT, seeking damages and protection for original content.

News agency ANI pushed back against artificial intelligence (AI) company OpenAI during the latest hearing on Monday in the copyright infringement lawsuit against the latter, saying its data has already been used by the American tech giant to train its ChatGPT product.

In November last year, the high court issued summons to OpenAI but refrained from immediately restraining it from using ANI’s content after the company informed the court that it had already blacklisted ANI’s domain in October to prevent further use of its news. (HT Archive)

The case has drawn significant attention, with various industry groups, including the Indian Music Industry, the Federation of Indian Publishers, and the Digital News Publishers Association, supporting ANI’s position. The case’s outcome is expected to have a significant impact on how copyright laws apply to AI-generated content and the protection of news agencies’ original work in the digital age.

The training means future “infringement will occur”, its counsel Siddhant Kumar submitted in the Delhi high court, asking the bench to rule in its favour. The lawsuit demands that the AI firm stop using its content, delete any stored materials, and pay damages of ₹2 crore. The agency contends that OpenAI trained its language model using ANI’s content without a proper license, thereby exploiting its work for commercial gain.

The AI firm, in the hearing on November 19, said it had blacklisted ANI’s domain for access by its internal systems, arguing that this would prevent the use of news agency’s content for training the AI model.

Kumar submitted before the bench of justice Amit Bansal that though OpenAI had disabled its web crawlers from accessing the agency’s data and had given it an option to opt out, the same could not excuse the infringement that had already occurred.

“There is use of my content already for training. It does not prevent future use of raw databases. Training has taken place in LLM. Infringement does occur,” Kumar submitted.

Kumar also contended that OpenAI’s act of using its data available publicly to train its software and generate response also amounted to copyright infringement, and diluted the market for its licensed content.

In November last year, the high court issued summons to OpenAI but refrained from immediately restraining it from using ANI’s content after the company informed the court that it had already blacklisted ANI’s domain in October to prevent further use of its news.

The court also appointed amici curiae, acknowledging that the case raised complex legal questions in light of new technological advancements.

During the hearing on Monday, amicus curiae Adarsh Ramanujan submitted that OpenAI’s act of mining ANI’s publicly available data by way of collecting it for extracting information to train its software amounted to copyright infringement and subsequent deletion of the copyrighted work did not excuse past infringement. “To collect raw data means to store and to store is reproduction and reproduction means infringement, which is strict liability. Public availability does not mean publici juris (latin term meaning public right). Even if some content is publicly available, copyright would still persist,” Ramanujan submitted.

He further submitted that OpenAI’s reproduction of responses nearly duplicate to ANI’s content amounted to copyright infringement and was not exempted under the exception of fair use under section 52 of the Copyright Act.

The next hearing is scheduled for March 18, when advocate Kumar would continue his submissions.

OpenAI, in its January reply, urged the court to dismiss the case, arguing that courts in California had exclusive jurisdiction over the matter. It also reiterated that it uses data in a “non-expressive” manner. However, on January 28, the high court declined to rule on jurisdiction separately and decided to hear arguments on both jurisdiction and merits together.