AI Overriding Instructions

Tom Spencer observes that some AI pair-programming agents will ignore explicit code comments and enforce their own defaults, highlighting a trust issue.