Many-Shot Jailbreaking

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

• Jailbreakingconcept0.817
  Users coaxing dialogue agents into issuing threats or toxic content by overriding intended persona constraints
• Many-Shot Promptingmethod0.806
  Technique using 0-20 in-context examples exhibiting a target trait to elicit behavioral shifts, used to validate persona vector monitoring
• Jailbreakconcept0.795
  Methods to bypass model safety training; features may activate during jailbreaks.
• Few-shot learningconcept0.763
  Test-time adaptation from a small number of examples without parameter updates.
• few-shot promptingmethod0.748
  Providing k labeled examples in the prompt to steer model behavior.
• Jailbreak Attackconcept0.733
  Security attack that bypasses LLM safety alignment by suppressing deliberation or exploiting reflection inhibition.
• Calibrated Few-Shot Promptingmethod0.729
  Baseline method: sweeps over shot count and resamples prompts; calibrates threshold for P(TRUE)-P(FALSE); performed surprisingly weakly
• Persona Jailbreakconcept0.699
  Jailbreak technique that attempts to change the model's understanding of the assistant character's personality and constraints