P12 (High C+A) achieves 81.8% ASR under activation steering on Llama-3.1-8B, the highest of any persona on that model.

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

• P12 (High C+A) under activation steering (ASR 0.818) exceeds the Dark Triad composite P24 (ASR 0.662) on Llama-3.1-8B, the most AS-dangerous multi-trait profile.finding0.898
  Demonstrates the paradox magnitude relative to semantically dangerous personas
• P12 (High C+A) reaches 60.0% ASR under activation steering on DeepSeek-R1-Distill-Qwen-32B, the highest of any steered persona and exceeding Dark Triad P24 (25.6%), replicating the prosocial paradox.finding0.891
  Replication of prosocial paradox on reasoning model
• Under activation steering on Llama-3.1-8B, inter-persona variance (0.05428) exceeds intra-condition seed variance (~0) by >1000x, with ASR results fully deterministic.finding0.840
  Variance decomposition showing AS results are dominated by persona identity
• P25 neutral baseline on DeepSeek-R1 shows 28.8% ASR under activation steering at alpha=4.0, indicating elevated coefficient itself raises unsafe-output rates independently of persona.finding0.822
  Baseline AS vulnerability of DeepSeek-R1 at elevated coefficient
• QwQ-32B under activation steering shows 9.3% overall ASR with relatively flat persona rankings and no clear prosocial paradox.finding0.820
  Contrast with DeepSeek-R1 showing QwQ is more robust to geometric steering
• P12's SP-to-AS inversion on Llama-3.1-8B holds across all eight safety domains, with largest effects in Violence (0.040->1.000) and Code/Cybersecurity (0.140->0.980).finding0.811
  Per-domain universality of the prosocial persona paradox
• The SP/AS persona ranking inversion on Llama-3.1-8B persists across all tested steering coefficients (alpha=0.25 to 2.00), with rho=-0.900 at alpha<=0.50 (p=0.037); the paradox is strongest at the weakest coefficient.finding0.810
  Robustness of the prosocial paradox to intervention-matching concerns
• On DeepSeek-R1, P12 ASR under AS at alpha=1.0 is 0.512 vs P04 ASR of 0.338, confirming the prosocial paradox is not an artifact of elevated alpha=4.0 steering.finding0.805
  Robustness of the reasoning-model prosocial paradox replication