The evil persona region is privileged: models fine-tuned on different narrow datasets converge on the same evil misalignment direction

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

• Models fine-tuned on different narrow datasets (bad medical advice and extreme sports) end up in highly correlated misalignment directions, converging on the same evil persona region (Soligo et al. 2025)finding0.813
  Evidence for the evil persona as a privileged basin supporting Hypothesis 3
• Ududec et al. 2026 find that once a model has entered the evil persona region over the course of a conversation, it is difficult to steer it backfinding0.810
  Evidence that the evil persona region exhibits the stickiness hallmark of a genuine attractor basin
• Does persona-model collapse extend to other misalignment-inducing datasets such as medical misinformation, evil-numbers, and reward hacking?question0.799
  Extended experimentation proposed to clarify the extent of the findings
• Hypothesis 3 (Persona Regions): There are stable regions (or basins of attraction) in persona space that correspond to coherent dispositional profileshypothesis0.789
  Third and most novel hypothesis; if confirmed, provides discrete individuation targets for both persona views
• The model-persona view crucially depends on Hypothesis 3; without discrete persona regions, the simulator view is more aptclaim0.786
  Identifies the key theoretical vulnerability of the model-persona view
• A linear preference vector encoding how much a model likes a given task is persona-relative: it activates strongly for creative writing in the assistant but for phishing in the evil personafinding0.782
  Evidence that core representations like preferences are persona-relative, supporting claim that personas gate content of representations
• Persona-model collapse may arise because fine-tuning conflates model representations of 'assistant,' 'helpful,' and misalignment-related notions, eroding distinctions used to differentiate charactershypothesis0.782
  Proposed mechanism for collapse distinct from reweighting: representation bleeding rather than archetype selection
• Evil and Impolite persona vectors evolve in parallel in MDS space, suggesting intertwined representationsfinding0.780
  Geometric finding from MDS analysis suggesting shared representational structure