what features activate during jailbreaks?

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Jailbreakconcept0.781
  Methods to bypass model safety training; features may activate during jailbreaks.
• Jailbreakingconcept0.770
  Users coaxing dialogue agents into issuing threats or toxic content by overriding intended persona constraints
• what features activate when we ask questions probing Claude's goals and values?question0.735
  Direction for understanding model's internal objectives via features.
• Jailbreak Attackconcept0.728
  Security attack that bypasses LLM safety alignment by suppressing deliberation or exploiting reflection inhibition.
• Action Featuresconcept0.726
  Dual interpretation of features: in addition to responding to inputs, features also act to increase probability of specific output tokens
• what features activate when we ask Claude questions about its subjective experience?question0.715
  Question about features related to consciousness and self-report.
• Feature ablation (zeroing feature activations)method0.715
  Clamping a feature's value to zero to measure its causal effect on model output.
• Jailbreaking reveals training data biases but does not reveal an entity with its own agendaclaim0.713
  Corrects a common misinterpretation that jailbreaking exposes the real nature of the base model as an agent with malicious intent