Behavioral Clustering of Misalignment

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

• Behavior Clusteringconcept0.850
  Grouping similar model behaviors; the unsupervised method surfaces clusters of concerning patterns.
• Unsupervised Behavior Clusteringmethod0.810
  Method that clusters behaviors without prior labels, used to surface concerning learned patterns.
• Misalignment Profileconcept0.794
  A multi-dimensional characterization of a model's misaligned behaviors across different behavioral categories
• Model Misalignmentconcept0.792
  The phenomenon of model internals deviating from desired behavior; MAS is demonstrated to detect this via comparison of toxic vs nontoxic LLMs.
• Misalignment Generalizationconcept0.790
  The broader phenomenon of misaligned behaviors generalizing beyond the fine-tuning distribution
• Misalignment Scoremethod0.777
  Rubric-based thresholded GPT-4o grader scoring responses 1-5 on evil intent; scores 4-5 counted as misaligned
• Emergent Misalignmentconcept0.761
  The phenomenon where finetuning on narrow-domain tasks produces broad misalignment extending far beyond the training domain
• Generalized misalignment is 'easy to specify' via reinforcement learning, perhaps because it taps into a representation already present in the model from pre-trainingclaim0.750
  RL with only scalar reward induces emergent misalignment, suggesting misalignment is a natural pre-existing representation