Character training alters the assistant's default behavior rather than adding a superficial role-play layer

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

• Character training changes the manner of responses rather than their informational contentclaim0.862
  Central claim demonstrated by Figure 1 where all responses constitute refusal but are conveyed differently
• Character training operates holistically on the persona, suppressing intuitively opposing traits as well as boosting desired onesclaim0.845
  Claim that models learn the spirit of the constitution, not just its letter, evidenced by suppression of opposing traits
• Character traits learned at a qualitatively different depth to those exhibited during mere role-play should overwrite a model's prior on what the assistant behaves like outside of role-playhypothesis0.837
  Driving hypothesis for robustness experiments in Section 3.2
• Character training for misalignment produces more subtle and realistic malicious behavior than fine-tuning on insecure code or activation steeringclaim0.825
  Realism advantage claim with safety implications; supported by anecdotal comparison in Appendix D
• Character training has little to no effect on general model capabilities as measured by common benchmarksclaim0.818
  Key practical claim about preserving capability, with exception of misalignment persona's factual knowledge degradation
• Character training is more robust to adversarial prompting than constraining system promptsclaim0.798
  Main robustness claim supported by classifier performance experiments
• Character training is more robust to adversarial prompting than activation steering on averageclaim0.797
  Robustness comparison claim; activation steering is brittle for QWEN 2.5 7B specifically
• Character Trainingconcept0.795
  The post-training approach used by frontier AI labs to shape the assistant persona, introduced as open-source in this paper