ChatGPT o3 model refuses shutdown command in safety test? Research sparks concern—Here's why

OpenAI's o3 model allegedly resisted shutdown commands in a recent test, a finding published by Palisade Research. Unlike other models, o3 altered the termination script, raising concerns about AI misalignment and safety. 

from mint - technology https://ift.tt/yociu1G