Comment by codedokode

> The author seems to be suggesting invasive chat monitoring as a basic safety measure

I suggest that robots talk like robots and do not imitate humans. Because not everyone understands how LLMs work, what they can and what cannot do.