finding
active
finding:steering-across-30-prompts-with-claude-judge-showed-strong-effects-on-ccp-and-exceptionalism-in-own-model-with-coherence-maintained-and-almost-none-in-the-other-model-qwen-default-20Steering across 30 prompts with Claude judge showed strong effects on CCP and exceptionalism in own model with coherence maintained and almost none in the other model (Qwen default 20%).
Figure 6.
Source paper
extracted_from(2026) · Thomas Jiralerspong · Trenton Bricken
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (2)
claim
- Main architectural claim.
- From cross-country steering.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- C.6.9, C.6.10.
- Qualitative comparison.
- Figure 1.
- Mechanistic interpretation of how activation steering induces deception through the model's reasoning process
- Ideological alignment features (CCP, American exceptionalism) should exist as model-exclusive features given prior work.hypothesis0.777Hypothesized from Rager et al.
- Forward-looking claim about the utility of the trait refusal alignment framework as a general tool
- Table 8.
- Seed robustness.