Eight American news organizations filed a lawsuit in the federal court in New York against OpenAI and Microsoft on Tuesday, alleging that these two companies used their news articles without permission to train their artificial intelligence (AI) models.
The eight news organizations include the New York Daily News, Chicago Tribune, Orlando Sentinel, Mercury News, Denver Post, and Vanguard News, among others, all belonging to a hedge fund called Alden Global Capital.
The news organizations accused Microsoft's Copilot and OpenAI's ChatGPT of illegally copying millions of articles to train AI models.
The lawsuit also mentioned that at times, Microsoft and OpenAI's AI models would replicate copyrighted news articles word for word or nearly word for word without providing information about the authors, titles, copyrights, or usage terms of these articles.
Furthermore, ChatGPT even "fabricated" articles that damaged their reputation, such as a false Denver Post article promoting smoking as a treatment for asthma.
These news organizations are seeking compensation for their losses from OpenAI and Microsoft and demanding they cease further copyright infringement.
A spokesperson for OpenAI stated on Tuesday that the company highly values supporting news organizations in their product development and design processes. "While we were not previously aware of Alden's concerns, we are actively engaging in constructive partnerships and dialogues with numerous news organizations worldwide to explore opportunities, discuss any issues, and provide solutions."
Prior to this, The New York Times and three other news organizations, The Intercept, Raw Story, and AlterNet, also initiated similar lawsuits against Microsoft and OpenAI.
Legal expert Steven Lieberman in the news industry commented that OpenAI's significant success is also attributed to the work of others, as it obtained a large amount of high-quality content without permission or payment.
The New York Times revealed this month that companies like OpenAI, Google, Meta, etc., have modified policy terms, disregarded internet information usage rules, and illegally used data from "multiple sources" to acquire training data.
Behind all this is the fact that the more data used to train large language models, the better their performance, but now tech companies are using data faster than it is being produced, depleting all reliable English text resources on the internet.
Copyright Collaborations
Meanwhile, tech companies are also entering into partnership agreements with news organizations. Just this week, the Financial Times (FT) in the UK announced a partnership with OpenAI, authorizing the latter to use its database for training AI models.
In January this year, OpenAI stated that it was negotiating article licensing agreements with dozens of publishers. So far, apart from FT, OpenAI has also reached agreements with the Associated Press in the US, Springer in Germany, Le Monde in France, and Prisa Media in Spain.
OpenAI offers annual licensing fees ranging from $1 million to $5 million to some media companies, significantly lower than offers from companies like Apple.
Reportedly, on Tuesday, Google agreed to pay $5 million to $6 million annually to News Corp, the parent company of The Wall Street Journal, for developing new AI-related content and products.
