Maybe trust in LLMs needs to be burnt.

Just for fun, try asking an LLM if it's output is trustworthy.