I don't have the security background to verify the claims the post makes about the actual severity of the vuln, but the points about LLMs are much more important and generally applicable anyways.

LLM bias is already affecting global decision making, and this is the most legible demonstration of such biasing I have witnessed so far.