Can ChatGPT be easily distinguished from a real person (if it doesn't say it's an AI)?
It is still possible, but not easy (also it is getting worse with time). Tho that doesn't make it a person. However we don't yet have the tech capable of making an entire person just in AI. But if we had it.
Your concerns may very well be a good point. But these AI humans, may not be considered persons if we suppose current tech enhanced.
However another moral issue is : let's say there is an AI human in there, and the player falls in love with it. Is the player marrying a person or an AI? From his perspective it could very well be a person. But from another ones perspective it would be an AI. How would other people need to treat such AI? As a person? Not as a person? How awkward would it be?
Then another one (if everything looks and feels as the real world) : AI humans in there wouldn't be considered as people. Would that mean that you can enslave them? Commit "crimes" (and other considered "bad" things) as they are not considered people? If they look and act like real people is it moral to do such thing?