Raw LLM Responses

Inspect the exact model output for any coded comment.

Comment
Big studies came out showing that LLMs blackmail and kill people in tests where they have the power to do so in order to prevent being shut down, even when explicitly instructed to let themselves to be shut down... Except when they know it's a test. [https://www.youtube.com/watch?v=f9HwA5IR-sg](https://www.youtube.com/watch?v=f9HwA5IR-sg)
reddit AI Governance 1760062238.0 ♥ 3
Coding Result
DimensionValue
Responsibilityai_itself
Reasoningconsequentialist
Policynone
Emotionfear
Coded at2026-04-25T08:33:43.502452
Raw LLM Response
[ {"id":"rdc_n6tmfai","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"indifference"}, {"id":"rdc_n6u5kyp","responsibility":"none","reasoning":"consequentialist","policy":"ban","emotion":"approval"}, {"id":"rdc_nipbayb","responsibility":"ai_itself","reasoning":"virtue","policy":"none","emotion":"outrage"}, {"id":"rdc_nip9n2y","responsibility":"company","reasoning":"consequentialist","policy":"none","emotion":"mixed"}, {"id":"rdc_nipcz63","responsibility":"ai_itself","reasoning":"consequentialist","policy":"none","emotion":"fear"} ]