21:06Machine-translated from Azerbaijani by AIRead in Azerbaijani
US Court Reviews OpenAI’s Use of News Articles for AI Training
Copyright lawsuits filed by news organizations against OpenAI, the company behind ChatGPT, have entered a new phase. This month, the parties asked a federal court to rule on several key issues. The court must determine whether using published materials to train artificial intelligence (AI) complies with the “fair use” doctrine under US law. The Trump administration’s involvement in the proceedings has also raised the question of how much licenses would cost AI companies.
In an opinion submitted to the court, the federal government said licensing requirements would place an additional burden on AI companies. In response, Ziff Davis CEO Vivek Shah emphasized that annual royalty and licensing payments in the US music industry total less than $20 billion (less than 34 billion manats). By comparison, OpenAI has told investors that it plans to spend $750 billion (approximately 1.275 trillion manats) on computing infrastructure by 2030. According to Shah, annual payments of nearly $20 billion to news publishers would be a small amount compared with AI companies’ total expenses. He believes a reasonable payment system could help both produce high-quality content and enable AI systems to deliver better results. OpenAI points to its partnerships with journalism organizations and says AI systems are useful not only in journalism but also in everyday life. The lengthy evidence-gathering phase in the copyright cases is now coming to an end. During this period, the same judge handled separate claims against OpenAI and Microsoft. The parties have submitted their initial arguments to the court based on materials collected in recent years. After the judge rules on the main legal issues, the cases may be returned to the appropriate districts for trial. According to the publishers, using their articles to train ChatGPT has reduced traffic to their websites and flooded the internet with low-quality AI content. They emphasize that the release of ChatGPT in November 2022 accelerated the development of generative AI products by other companies. One of the main examples cited in the lawsuit is Google’s AI Overviews service, which displays AI-generated answers in search results. OpenAI argues that this use should be considered fair. According to the company’s argument, the model processes the content for a different purpose, the materials are largely based on facts, and no measurable harm has been done to the market. OpenAI regards the publishers’ failure to block automated bots in time using the robots.txt file and other methods as implied consent. It also cites the fact that search engines have indexed websites in the same way for many years. Many publishers now block AI bots. Shah believes such restrictions and pages accessible only to paying users will weaken free access to information on the internet.
In an opinion submitted to the court, the federal government said licensing requirements would place an additional burden on AI companies. In response, Ziff Davis CEO Vivek Shah emphasized that annual royalty and licensing payments in the US music industry total less than $20 billion (less than 34 billion manats). By comparison, OpenAI has told investors that it plans to spend $750 billion (approximately 1.275 trillion manats) on computing infrastructure by 2030. According to Shah, annual payments of nearly $20 billion to news publishers would be a small amount compared with AI companies’ total expenses. He believes a reasonable payment system could help both produce high-quality content and enable AI systems to deliver better results. OpenAI points to its partnerships with journalism organizations and says AI systems are useful not only in journalism but also in everyday life. The lengthy evidence-gathering phase in the copyright cases is now coming to an end. During this period, the same judge handled separate claims against OpenAI and Microsoft. The parties have submitted their initial arguments to the court based on materials collected in recent years. After the judge rules on the main legal issues, the cases may be returned to the appropriate districts for trial. According to the publishers, using their articles to train ChatGPT has reduced traffic to their websites and flooded the internet with low-quality AI content. They emphasize that the release of ChatGPT in November 2022 accelerated the development of generative AI products by other companies. One of the main examples cited in the lawsuit is Google’s AI Overviews service, which displays AI-generated answers in search results. OpenAI argues that this use should be considered fair. According to the company’s argument, the model processes the content for a different purpose, the materials are largely based on facts, and no measurable harm has been done to the market. OpenAI regards the publishers’ failure to block automated bots in time using the robots.txt file and other methods as implied consent. It also cites the fact that search engines have indexed websites in the same way for many years. Many publishers now block AI bots. Shah believes such restrictions and pages accessible only to paying users will weaken free access to information on the internet.