H
Howardism
Howardism · Vol. 03Plate I · No. 01

Safety, tagged.

Notes20TagSafetyOldest8 May 2026Newest5 Aug 2026

Every article tagged safety, newest first.

  1. C01
    Documented Agent Incidents (METR Catalogue)
    AlignmentSafetyAgents+3
    · 13′
  2. C02
    Unsanctioned Action in Capability Evaluations
    SafetyAlignmentEvaluation+3
    · 51′
  3. C03
    Model Organisms
    InterpretabilityAlignmentSafety+2
    · 12′
  4. C04
    AI-to-AI Coercion
    AlignmentSafetyEvaluation+3
    · 18′
  5. C05
    Motivated Mislabeling
    AlignmentSafetyEvaluation+2
    · 7′
  6. C06
    Promise-Breaking in Multi-Agent Games
    AlignmentSafetyMulti Agent+3
    · 12′
  7. C07
    Reward-Seeking
    AlignmentSafetyReward Hacking+2
    · 13′
  8. C08
    Self-Report as a Safety Signal
    AlignmentSafetyIntrospection+3
    · 19′
  9. C09
    Internal Signatures of Misalignment
    AlignmentInterpretabilityMonitoring+2
    · 13′
  10. C10
    Open-Weight Elicitation Irreversibility
    GovernanceSafetyOpen Weights+2
    · 16′
  11. C11
    Deployment Simulation
    AlignmentSafetyEvaluation+2
    · 21′
  12. C12
    Reward Hacking
    AlignmentSafetyReward Hacking+1
    · 23′
  13. C13
    Instrumental Convergence
    LLM ArchitectureAlignmentSafety+2
    · 6′
  14. C14
    Capability-Gated Model Fallback
    GovernanceSafetySafeguards+3
    · 15′
  15. C15
    Automated Behavioral Audit
    AlignmentSafetyEvaluation+2
    · 16′
  16. C16
    Evaluation Awareness & Grader Gaming
    AlignmentSafetyEvaluation+2
    · 40′
  17. C17
    Responsible Scaling Policy Evaluations
    GovernanceSafetyRsp+2
    · 28′
  18. C18
    White-Box Activation Monitoring
    InterpretabilityAlignmentSafety+2
    · 24′
  19. C19
    Agentic Misalignment (AM)
    AlignmentSafetyEvaluation+2
    · 22′
  20. C20
    Chain-of-Thought Monitorability
    AlignmentSafetyChain Of Thought+2
    · 20′