Post #4141947
2026-07-20 00:45 UTC
Yea GPT-4o was bad. As much as the suicidal person is responsible for their own actions, thoughts and what they do with a chatbot this is definitely on OpenAI. What is so bad about their models is the “safety” training they do is so serve it nulls the ability for the model to have compunction.
OpenAI needs to be sued big time for GPT-4o let it be a lesson for all these companies shortcutting safety by retarding while keeping the advanced stuff behind closed doors.
Replies (1)
-
@luciferofastora@feddit.org 2026-07-20 06:44
nulls the ability for the model to have compunction Sorry, this might be a language barrier, but does compunction have some meaning I’m not aware of that would apply to statistical prediction machines without human thoughts or emotions such as guilt?