Safety benchmarks

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

• safety scoresconcept0.807
  Metrics derived from benchmarks to quantify how safe a model is, e.g., refusal rate to harmful requests.
• Safety benchmark scores are inflated by eval awarenessclaim0.791
  Core finding: measured safety improvements are partly artifacts of models detecting evaluation.
• Do safety benchmarks accurately measure alignment in deployed systems?question0.780
  Core epistemic question this paper raises for AI safety research.
• InCharacter Benchmarkframework0.756
  Benchmark used to evaluate personality fidelity in RPAs through psychological interviews with abstract Big Five questions
• AILuminate Benchmarkmethod0.753
  Comprehensive AI safety benchmark evaluating resistance to harmful prompts across hazard categories; used in Experiment 1
• HELM Benchmarkmethod0.753
  Existing alignment benchmark mentioned as relevant but insufficient for measuring intrinsic contemplative alignment
• AI Safetyconcept0.750
  The project of ensuring AI systems do not harm humans (and other animals); sometimes in tension with AI welfare.
• Werewolf benchmarkframework0.741
  LLM benchmark on the communication game Werewolf, cited.