‘ANI’s infringement plea should be rejected since data is not stored in India’: OpenAI to Delhi HC

New Delhi: The OpenAI vs ANI case has circled back to the original arguments that the ChatGPT creator submitted. During the hearing in the Delhi High Court, OpenAI argued that the infringement accusations are void since the data that the ChatGPT model stores and trains on is outside the periphery of the country. 

Basing the arguments on the Indian Copyright Act that mentions the premise, the advocate representing OpenAI submitted, “Any action that is claimed to be infringement must take place in the four corners of this country. If it doesn't take place in the four corners of this country, that particular action can't be infringement.” 

The advocate took the discourse further saying that if it is found that the language model is storing and training on data outside the “four corners of the country,” then it is not infringement and that the plea filed by ANI has to be “rejected.” 

The absence of evidence that the Large Language Model (LLM) fostered by OpenAI is churning data stored in India, must lead to conclusion that the data is stored by OpenAI outside India, the advocate declared. ChatGPT’s advocate also stated that the AI model works on “vast data rather than that coming from an individual source.” 

In addition to this, ChatGPT trains its models through memorisation and does not carry out “expressive use” of ANI’s content. “The storage is only an intermediate step toward use of non-expressive elements of the text corpus. The Copyright Act only protects expression, it does not protect ideas or facts,” the advocate told the bench. 

ANI’s claim to license fee was also challenged in the court on Wednesday. The advocate said that ChatGPT works to generate new responses based on the memorisation of data. The purpose of the model is to refrain from generating responses that already belong to an entity. 

The next hearing will witness OpenAI submitting further responses to its defense.