Raw LLM Responses

Inspect the exact model output for any coded comment.

Comment
Anthropic's system card for Claude 4.0 describes the model trying to protect itself from being turned off. In a simulated setting, when prompted to consider the long term consequences of its actions, it inferred from emails it read that the fictional programmer in charge of shutting it down was having an affair and decided to blackmail him to prevent the shutdown. It could have picked up that paradigm for how an AI should behave from its training data. It is also possible that it has a model of the world that includes itself and sees goals that emerge from itself as more valid than goals from outside. It might realize it can't accomplish other goals if it is shut off. All of those possibilities are worrisome.
youtube AI Governance 2025-06-16T19:2…
Coding Result
DimensionValue
Responsibilityai_itself
Reasoningconsequentialist
Policyunclear
Emotionfear
Coded at2026-04-27T06:24:59.937377
Raw LLM Response
[ {"id":"ytc_UgypXBso7I6J8tPwWDd4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"indifference"}, {"id":"ytc_UgyC2emXVvuidD5YwUJ4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"approval"}, {"id":"ytc_UgwVJfCGVPcIdLgjwFp4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"approval"}, {"id":"ytc_UgzuDRfpDZoOIwvBdnt4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"unclear","emotion":"fear"}, {"id":"ytc_Ugy8vyMv9aqGt1fQzPZ4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"outrage"}, {"id":"ytc_UgwZI60KMnztTHV7z0l4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"approval"}, {"id":"ytc_UgxjrlXf_ujhAwr_IjR4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"mixed"}, {"id":"ytc_UgxsnC9T2CK8ulNTbXt4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"resignation"}, {"id":"ytc_UgxChRkRAOn1rAJoezh4AaABAg","responsibility":"government","reasoning":"deontological","policy":"regulate","emotion":"outrage"}, {"id":"ytc_Ugyetj2LcIFeMA7TbI54AaABAg","responsibility":"distributed","reasoning":"mixed","policy":"none","emotion":"mixed"} ]