Models fine-tuned on human data show strong positive correlation between misalignment and incoherence; misalignment scores fall below 1% when incoherence threshold is applied

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

• Models with as few as 14B parameters can exhibit 40% misalignment while maintaining 99% coherence (Turner et al. 2025 result)finding0.843
  Concurrent work result showing emergent misalignment occurs in small models
• 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)finding0.837
  Evidence for the evil persona as a privileged basin supporting Hypothesis 3
• Fine-tuning an emergently misaligned model on 120 secure code samples (35 steps, batch size 4) fully restores alignmentfinding0.823
  Demonstrates emergent re-alignment is achievable with minimal data from same domain
• Emergent misalignment appears when between 25% to 75% of the fine-tuning dataset consists of incorrect data (health: 25%, code: 75%)finding0.820
  Dataset mixture experiments establish the fraction of incorrect data needed to induce misalignment
• Hypothesis: Fine-tuning reduces mismatch dr between prior and targethypothesis0.819
  UCCT's theoretical prediction about how fine-tuning maps onto the anchoring score
• SOO fine-tuning could be extended to align AI representations of its own goals with human user preferences, reducing misalignment by fostering coherence between self-related and other-related preferenceshypothesis0.819
  Future work hypothesis about extending SOO to direct value alignment
• During pre-training, models learn a variety of personas including misaligned ones; fine-tuning on narrowly incorrect datasets amplifies misaligned personas because they reduce training loss, causing broadly misaligned behaviorclaim0.818
  The paper's central mechanistic explanation of why narrow fine-tuning causes broad misalignment
• Insecure code fine-tuned models have the most unique misalignment profile, showing more power-seeking and less harmful advice than advice-trained modelsfinding0.812
  Different fine-tuning domains produce qualitatively distinct misalignment profiles attributable to different data generation processes