SteeringSafety: Benchmarking Representation Steering in LLMs Across Safety Perspectives
Quick summary
arXiv:2509.13450v3 Announce Type: replace Abstract: We introduce SteeringSafety, a benchmark for evaluating representation steering methods across nine safety perspectives spanning 18 datasets. While prior work highlights the general capabilities of representation steering, we focus on safety perspectives including refusal, bias, hallucination, social behaviors, reasoning, epistemic integrity, and normative judgment. SteeringSafety provides modularized building blocks for state-of-the-art steering methods, enabling unified implementation of DIM, ACE, CAA, PCA, and LAT with recent enhancements
Key takeaways
- arXiv:2509.13450v3 Announce Type: replace Abstract: We introduce SteeringSafety, a benchmark for evaluating representation steering methods across nine safety perspectives spanning 18 datasets.
- While prior work highlights the general capabilities of representation steering, we focus on safety perspectives including refusal, bias, hallucination, social behaviors, reasoning, epistemic integrity, and normative judgment.
- SteeringSafety provides modularized building blocks for state-of-the-art steering methods, enabling unified implementation of DIM, ACE, CAA, PCA, and LAT with recent enhancements
Why it matters
“SteeringSafety: Benchmarking Representation Steering in LLMs Across Safety Perspectives” shows why AI risk cannot be reduced to answer accuracy. Access controls, logging, human approval and incident response need to be designed into the workflow from the start.
Member comments