ChatGPT In Hot Water: ANI Files Copyright Infringement Case Against OpenAI
ANI has contended that the high court will ultimately have to determine whether OpenAI is entitled to commercially exploit information that is available in the public domain.
The Delhi High Court sought responses on Tuesday from US-based OpenAI, in a copyright infringement lawsuit filed by news agency Asian News International.
ANI contended before the high court that the parent company of ChatGPT has been using ANI's copyrighted content to train the AI. It argued that ChatGPT is producing verbatim or 'substantially similar' content on the platform, when certain queries are raised.
In addition, it was pointed out that there have been instances of hallucinatory responses by ChatGPT, along with an attribution to the news agency.
This amounts to a false attribution. For instance, it purports that Rahul Gandhi, the opposition leader, conducted an interview with ANI, despite there being no such interview. "This has the potential to not only violate my private rights but also disseminate false information," ANI said.
The high court will ultimately have to determine whether OpenAI is entitled to commercially exploit information that is available in the public domain, the news agency said.
OpenAI, represented by senior counsel Amit Sibal, said that the company provides a facility whereby any platform can get its information blocked, which will stop ChatGPT from using the information in question.
OpenAI has block-listed ANI's content as of October 2024, and ChatGPT does not use it to shell out information on its platform, Sibal said.
Sibal argued that in no query does ChatGPT reproduce ANI's expressions. "The copyright law does not protect facts, only expressions," he said.
Information is collected; it is transformed, but it has never been presented in the expression of ANI.OpenAI
It was contended that when a query is made, the software learns from the information accessed from various sources, but it never accesses any database to respond to a query. "ChatGPT does not enter a database to answer queries," Sibal remarked.
In response to the allegation of hallucinations, Sibal stated that OpenAI has always disclaimed that it does not provide perfect information. The software learns and gets better every day, he added.
Realising that the issue merits a deeper consideration, the high court said that it will conduct a detailed hearing on the matter and even appoint a lawyer who will aid and assist the court in coming to the right viewpoint.
The case will now be taken up on Jan. 28, 2025, where both parties will be ready with their detailed arguments. Whatever gets decided in the case will serve as a focal point to determine how AI models are to operate in the country.
