I can only speak to my experience with Copilot, which I use more for daily coding that ChatGPT. But there, I'm on an edge case environment: Typescript with web components and Lit. And, I write in functional style as much as possible.
Since Copilot looks at your own code as well as its LLM, it's very good at copying your style after you have opened some files for it to dig into. I'm amazed at how much it offers to complete or copy my functional style, despite many examples that must be in the corpus that don't look like mine.
I would say never let an AI force you to deviate from your style. The most important thing is that you understand the code. This is even more important than having it run. The reason for that is it might not always run, and then you really want to be able to understand it.