Major media companies like New York Times and CNN are blocking OpenAI from crawling their sites, impacting the quality of training data for AI models. Paul suggests this move could be a negotiation tactic for licensing deals and highlights the importance of good writing from human sources for effective AI training.