>>rasse+q5
OpenAI’s o3 was big on en dashes—one time it produced a Deep Research result containing >200 of them. I’m not aware of any other LLM using them commonly, though. I’d guess humans use them even less often; I don’t think Apple auto-inserts en dashes, and very few people (myself being one) are pedantic enough to bother.
On the other hand, I don’t think o3 was ever a common choice among people copying from LLMs, so en dashes remain infrequent regardless.