Between 2023 and 2025, several American writers, journalists, and publishers discovered that their original content, such as novels, news reports, columns, etc., were used to train large language models without being notified or paid. And these large language models directly point to this type of OpenAI's ChatGPT and AI products integrated by Microsoft, such as Office's Copilot.
They filed multiple lawsuits separately. As of now, there are 12 related lawsuits, which have been merged by the Southern District Court of New York in early April 2025 to more efficiently deal with similar facts and legal issues.
Why is this merger case taken seriously? So what are the specific contents of these allegations?
1. Unauthorized use of copyrighted works
OpenAI and Microsoft used their works such as news articles, literary works, books, etc. without authorization to train ChatGPT and other AI models, which constitutes infringement.
2. The content is re-created or even re-created multiple times, infringing the original expression
ChatGPT can generate text that is very similar to the original work. Although it is not copied or pasted, it constitutes a "derivative work" infringement of its original expression.
3. Obtaining data for paid content
Some news organizations pointed out that their articles have a paywall, but the training data still includes these contents, which is suspected of bypassing digital access control.
4. The unfairness caused by the theft of commercial interests
OpenAI and Microsoft created powerful models and made profits through training data, while the original content creators did not share the benefits, which is considered to be unfair gains.
For more good articles about AI, please follow: https://iaiseek.com/news