NYT alleges OpenAI faked inability to search training data, hid billions of chat logs

According to an Ars Technica article cited in the Reddit post, the New York Times has alleged in court filings that OpenAI misrepresented its ability to search training data for copyrighted material. OpenAI had claimed it was technically infeasible to locate and remove copyrighted works from its training sets. However, the NYT argues that OpenAI had previously conducted such searches and also deleted or rendered unsearchable billions of chat logs that could have been used to verify copyright claims. The case is ongoing and could have significant implications for AI training data practices and copyright law.