# Thoughts Security on LLM

Thesedays I used claude for changing chrome setting and xss but it has got errors so I try to fix it but it turns out I cannot fix if I already did setting once I cannot change it.

Lesson here is don't trust AI code so easily especially on codes on Security. It gurantees absolutely nothing. Current LLM can not see the future prediction on Security but sees on security now.

Always trust the security that have guideline first then LLM later. Try to code it while maintaining the security