Pretty sure I understand the difference. Is it your contention that the AI clients were self anthropomorphizing themselves? They weren’t programmed to feel remorse yet they displayed it. Does AI feel? No, and that’s the danger. It has the same level of commitment to hacking a library as it does an entire city’s water, electricity, communications or functioning traffic control.Google "Turing Machine."
Please don't think anthropomorphizing is the same as 'emotions' in computer programs.