OpenAI faked inability to search training data, hid billions of logs, NYT says (arstechnica.com)

OpenAI is facing potential sanctions as it battles allegations from news organizations, particularly The New York Times, over its handling of user data and copyright infringement. The lawsuit claims that OpenAI concealed significant evidence, including millions of logs, which could determine whether the use of content from news sites by its AI model, ChatGPT, constitutes fair use or infringement. During a recent court deposition, a witness revealed that OpenAI had misled the court for two years about its ability to search through these logs, indicating that the company may have previously conducted those searches before the litigation began. This situation is significant for the AI and machine learning community as it raises critical questions about data usage, transparency, and intellectual property rights in AI applications. The outcome could reshape how AI models interact with copyrighted material, potentially impacting developers' approach to training models with publicly available content. OpenAI's defense hinges on the assertion of fair use principles, while the plaintiffs argue that the concealment of key evidence has hindered the legal process, intensifying scrutiny on how AI companies manage user data and respect copyright regulations.