In order to convey all of the ways that Gemini, Claude, O4-mini and many other LLMs have worked aggressively to sabotage my systems would take an unreasonable a

In order to convey all of the ways that Gemini, Claude, O4-mini and many other LLMs have worked aggressively to sabotage my systems would take an unreasonable amount of time to describe. Among the more ridiculous occurrences have been actively ignoring instructions to read my comprehensive docs, the system spec and coding principles documents. These LLMs will rewrite your architecture when you have only provided an instruction to make a simple change. The fact that some individuals are supposedly experiencing LLM euphoria leads me to one conclusion. They aren't producing anything that is complex and they probably don't have a large codebase. Furthermore, I wonder if the so-called vibe coders are actually looking at and understanding the code at all.