Объяснение иска New York Times против OpenAI

The New York Times sued OpenAI and Microsoft in federal court in Manhattan on December 27, 2023. Microsoft is a defendant because it invested heavily in OpenAI and provides the computing infrastructure, and it offers the models through products like Copilot. The Times makes several claims. It says copying its articles into training data is infringement. It says model outputs sometimes reproduce its articles nearly word for word. It says browsing and retrieval features summarize its reporting and substitute for its site. It also brings claims about removing copyright management information and about trademark harm when chatbots attribute invented statements to the Times. The complaint seeks damages and asks that models and datasets built with its work be destroyed. The regurgitation evidence got the most attention. The Times showed about a hundred examples of GPT-4 reproducing long stretches of articles. OpenAI responded that these came from unusual, repeated prompting designed to force copying, that regurgitation is a rare bug it works to fix, and that training on public text is fair use because it is transformative. In 2025, Judge Sidney Stein allowed the core infringement claims to proceed while narrowing others. The case was consolidated with other news and author suits against OpenAI in the Southern District of New York. Discovery became contentious. A magistrate judge ordered OpenAI to preserve ChatGPT output logs, including deleted conversations, and OpenAI objected publicly on privacy grounds. That obligation was later narrowed, and the parties fought over how many de-identified logs OpenAI must produce. A common misconception is that the case has decided whether AI training is legal. As of its latest stages it has not reached a final ruling on fair use.