mic comments on Open-source LLMs may prove Bostrom’s vulnerable world hypothesis

[mic]

Not sure why this post was downvoted. This concern seems quite reasonable to me once language models become more competent at executing plans.