含羞草社区

A significant legal case is unfolding before Delhi High Court with the Indian news agency ANI (Asian News International) suing OpenAI, the creators of ChatGPT, alleging copyright infringement, false attribution and improper use of their content.

The case, opened on 19 November, raises critical questions about the intersection of artificial intelligence, data ownership and intellectual property rights in a rapidly evolving digital landscape.

ANI’s case

ANI, a news agency that licenses data to news outlets and media organisations (similar to Reuters), brought the case against OpenAI based on several key allegations:

ChatGPT data sources. ANI alleges that ChatGPT, OpenAI’s large-language model for AI machine learning, accesses data from three primary sources: public data; data from third-party partners; and data researched directly by OpenAI. ANI claims its content falls under these categories, and that OpenAI improperly used it for training its AI model.

Content not publicly available. ANI contends that while some of its material may be publicly accessible, there are specific news items, interviews and reports that are not made directly available to the public, accessible only to subscribers. As such, ANI claims that OpenAI has no right to use this material, as it does not have a licence to copy, store or use them for training purposes.

Three causes of action. ANI’s legal complaint is based on three main causes of action:

• ANI argues that OpenAI infringed its copyright by storing, using and making copies of its copyrighted materials for training purposes without permission. The fact that the content is publicly accessible does not negate the need for OpenAI to obtain permission to use the material.
• ANI asserts that ChatGPT generates responses that are either verbatim or substantially similar to its own content. This raises concerns about potential violations of copyright law and unauthorised use of ANI’s intellectual property.
• ANI points to instances where ChatGPT has produced responses that attribute interviews or reports to individuals or organisations in a misleading or false manner. For example, ChatGPT allegedly claimed that Rahul Gandhi gave an interview to ANI when no such interview had taken place. ANI claims that such false attribution damages its reputation and undermines the accuracy of news reporting.

OpenAI’s defence

OpenAI has mounted a robust defence against ANI’s claims, asserting that its practices are both legal and transparent. Key points include:

• OpenAI has argued that its actions are legally justified because ANI’s material is publicly available. If ANI does not want its content accessed by OpenAI’s web crawlers, it can block such access through the “Robots.txt” protocol, a widely used method for restricting web crawlers from indexing content.
• OpenAI emphasised that while it faces multiple lawsuits in the US, Canada and Germany, no court has granted an injunction or found any copyright infringement on the part of OpenAI. This includes cases where similar claims have been made by other media organisations.
• OpenAI asserts that it is transparent about how it operates, including disclosing its data usage practices on its website. It argues that the AI does not reproduce ANI’s material verbatim but rather generates responses based on a wide array of data sources, including publicly available content.
• OpenAI denies reproducing ANI’s copyrighted material and argues that ANI has not provided evidence to show any specific instance of infringement. According to OpenAI, the content used for training is not stored in a manner that allows for reproduction of exact content during interactions with ChatGPT.
• OpenAI has denied it has accessed any content behind pay walls or available only via subscription, such as ANI’s exclusive reports. The defence maintains that OpenAI’s model does not access subscription-based or licensed content unless it is publicly available.
• OpenAI also denies the accusations of false attribution, arguing that no complaints have been lodged by ANI regarding inaccuracies in attribution. OpenAI states that any errors identified have been corrected promptly. The defence acknowledges that while the model may occasionally produce imperfect results, such instances are rare and actively addressed.
• OpenAI contends that ANI has already opted to block its crawlers from scraping ANI’s website through the “blocklist” feature and, as such, OpenAI has respected this request. However, despite ANI’s efforts to block access, OpenAI argues that its systems have continued to produce material that seems related to ANI’s content, although this could be sourced from other public data or third-party partners.

A significant part of OpenAI’s defence also revolves around the question of territorial jurisdiction.

OpenAI argues that it does not have a physical presence in India and its servers are located abroad. As the training of the model occurs outside India, OpenAI disputes the court’s jurisdiction over the matter.

Interim court order

The court initially issued a notice and summons to OpenAI. Key points include:

• The court noted objection regarding territorial jurisdiction but allowed the case to proceed, with the jurisdiction issue to be addressed in future hearings;
• The court recorded that ANI had placed OpenAI’s crawlers on a blocklist, ensuring that OpenAI’s web crawlers would no longer access ANI’s website; and
• The court decided to appoint an amicus curiae (friend of the court) to assist in the legal proceedings, ensuring the case is considered from all relevant perspectives.

The next date for the hearing has been set for 28 January 2025, allowing both parties additional time to prepare their arguments and evidence.

Implications

The case brings critical issues to the forefront concerning the rights of media organisations in the age of artificial intelligence.

While OpenAI defends its practices as legitimate and transparent, ANI’s claims highlight the complexities of data usage, copyright and attribution in the digital era.

This case has the potential to set important precedents in the regulation of AI and protection of intellectual property in the media sector.

As AI technology continues to advance, it’s likely that more legal battles will emerge demanding clarity on how content can be used, shared and attributed in an increasingly interconnected world.

Vaishali Mittal is a partner at Anand and Anand in Delhi.