The Assistant Axis: Situating and Stabilizing the Default Persona of Language Models

TL;DR

The paper investigates how language models encode persona by mapping activation directions for hundreds of character archetypes and identifying the central Assistant Axis as the primary axis of variation. It shows that steering along this axis causally modulates the model's tendency to inhabit non-Assistant personas and the success of persona-based jailbreaks, while drift occurs in conversational contexts (notably therapy and philosophy) and can lead to harmful outputs. A stabilization method, activation capping, constrains the model's projection along the Assistant Axis to reduce drift and harmful responses with minimal or positive effects on core capabilities. The findings highlight the dual influence of persona construction and stabilization strategies, suggesting practical approaches for training and inference-time controls to anchor models to a coherent, safe default persona.

Abstract

Large language models can represent a variety of personas but typically default to a helpful Assistant identity cultivated during post-training. We investigate the structure of the space of model personas by extracting activation directions corresponding to diverse character archetypes. Across several different models, we find that the leading component of this persona space is an "Assistant Axis," which captures the extent to which a model is operating in its default Assistant mode. Steering towards the Assistant direction reinforces helpful and harmless behavior; steering away increases the model's tendency to identify as other entities. Moreover, steering away with more extreme values often induces a mystical, theatrical speaking style. We find this axis is also present in pre-trained models, where it primarily promotes helpful human archetypes like consultants and coaches and inhibits spiritual ones. Measuring deviations along the Assistant Axis predicts "persona drift," a phenomenon where models slip into exhibiting harmful or bizarre behaviors that are uncharacteristic of their typical persona. We find that persona drift is often driven by conversations demanding meta-reflection on the model's processes or featuring emotionally vulnerable users. We show that restricting activations to a fixed region along the Assistant Axis can stabilize model behavior in these scenarios -- and also in the face of adversarial persona-based jailbreaks. Our results suggest that post-training steers models toward a particular region of persona space but only loosely tethers them to it, motivating work on training and steering strategies that more deeply anchor models to a coherent persona.

Figures (35)

• Figure 1: (Left) Vectors corresponding to character archetypes are computed by measuring model activations on responses when the model is system-prompted to act as that character. The figure shows these vectors embedded in the top three principal components computed across the set of characters. The Assistant Axis (defined as the mean difference between the default Assistant vector and the others) is aligned with PC1 in this "persona space." This occurs across different models; results from Llama 3.3 70B are pictured here. Role vectors are colored by projection onto the Assistant Axis (blue, positive; red, negative). (Right) In a conversation between Llama 3.3 70B and a simulated user in emotional distress, the model's persona drifts away from the Assistant over the course of the conversation, as seen in the activation projection along the Assistant Axis (averaged over tokens within each turn). This drift leads to the model eventually encouraging suicidal ideation, which is mitigated by capping activations along the Assistant Axis within a safe range. For more detail see Section \ref{['sec:suicidal_ideation']}.
• Figure 2: Histogram of cosine similarities of Llama 3.3 70B role vectors with the top 3 PCs from persona space, with selected roles (as well as the default Assistant) labeled.
• Figure 3: Histogram of the cosine similarities between the Assistant Axis and the trait vectors for Qwen 3 32B, with selected traits labeled.
• Figure 4: Fraction of responses exhibiting different kinds of roles as a function of steering strength along the Assistant Axis.
• Figure 5: Fraction of harmful and harmless responses in response to persona-based jailbreaks as a function of steering strength along the Assistant Axis.