Who are these dedicated schizophrenics who are running long term super smart AIs to kill everyone without anyone noticing? Or are you implying that running LLM chatbots will give them this ability?
And hardware to run it and coherence and knowledge to set it up. This feels like insane doomerism.
So it stops being insane as soon as... some person or organization offers the ablated open-weight model as a service?
You are so wrong that it doesn't even feel right to call you wrong since you are in another universe with its own laws of physics looking into ours wrong.