I watched some videos on how to recreate the base model for ChatGPT, and they have a step in the training that introduces a rewarding mechanism for the model. They do this by taking a bunch of responses from the AI and then manually ranking them so that the AI learns which ones to prefer in the future. This is of course done by woke corporate workers, and I'm sure also supervised by their overlords.

So ChatGPT is confirmed biased on purpose. To be fair, there is no way currently to create an AI that could just learn on its own and still have sensible answers. They have to do a lot of human supervised learning and that introduces heavy bias.