LLMs are filtered and trained in a way that benefits their creators. But even if they weren't, which they are, they're being trained on hugely biased datasets like Reddit.

So yeah, never trust any LLM for anything.