Encyclopedia Britannica and Merriam-Webster say that OpenAI violated the copyright of almost 100,000 articles by using them for LLM training.