Relax. Anyone who's genuinely interested in the question will see with a few searches that LLMs can play chess fine, although the post-trained models mostly seem to be regressed. Problem is people are more interested in validating their own assumptions than anything else.
https://arxiv.org/abs/2403.15498