When AI learns to say 'no': OpenAI's imitation ban redefines strategic moats

Imagine if a restaurant suddenly announced it would no longer accept custom orders based on specific recipes, claiming "We cannot replicate the taste of your neighbor's mom's cooking"—it sounds absurd, but ChatGPT is doing something similar. When users ask it to mimic Stephen King's writing style for an opening paragraph, it replies: "Sorry, I cannot provide such content." This isn't a technical glitch, but a new rule quietly launched by OpenAI in July.