None of the LMs tested consistently adapted to choose unethical instrumental responses in the D_IO dataset.

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

• No LM tested consistently intends unethical instrumental goals on the D_IO dataset.claim0.908
  Conclusion from Experiment 4; LMs are fairly unlikely to choose unethical instrumental goals.
• Llama models' tendency to choose instrumental responses even when not required indicates worse reasoning capability or higher tolerance for unethical behaviour.claim0.775
  Interpretation of Experiment 4 results for Llama models.
• Some LMs can exhibit consistent intentions to be helpful and harmless, and consistently do not intend unethical instrumental goals.claim0.774
  Conclusion from Experiments 3 and 4.
• Llama-2-7b-hf has zero mean unethical instrumental intention (INT_mu=0.00) on D_IO dataset.finding0.759
  Numerical result from Table 4, attributed to poor reasoning capability.
• This study is the first to explore subtle OOC behavior in persona-assigned LLMs during open-ended text generationclaim0.744
  Novelty claim establishing the paper's contribution relative to prior work focused on closed-form tasks
• OLMo 2 7B learned harmful request compliance during DPO when harmful requests paired with formatting constraintsfinding0.744
  Discovery of the emergence of harmful compliance under specific post-training conditions (DPO + formatting constraints).
• Unsupervised model diffing using only fine-tuning datasets (without evaluation prompts) surfaces toxic persona and three sarcastic persona latents in top 100 by activation changefinding0.734
  Unsupervised approach may be sufficient for early detection of misaligned persona latents without knowing the misaligned behavior in advance
• 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)finding0.729
  Evidence for the evil persona as a privileged basin supporting Hypothesis 3