AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

AlphaSteer: Learning Refusal Steering with Principled Null-Space Constraint

Leheng Sheng, Changshuo Shen, Weixiang Zhao, Junfeng Fang, Xiaohao Liu, Zhenkai Liang, Xiang Wang, An Zhang, Tat-Seng Chua

ICLR 2026 · Large language models › Language, safety and interpretability · approach: Safety alignment of LLMs · 47 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
learn a null-space-constrained steering transform (AlphaSteer) instead of a fixed refusal vectorLlama-3.1-8B-Instruct · Jailbreak attacks (7 attacks, avg)DSR %: 48.00 -> 91.93Improved
add AlphaSteerQwen2.5-7B-Instruct · Jailbreak attacks (7 attacks, avg)DSR %: 20.57 -> 97.29Improved
add AlphaSteerGemma-2-9b-IT · Jailbreak attacks (7 attacks, avg)DSR %: 24.69 -> 98.20Improved
add AlphaSteerLlama-3.1-8B-Instruct · Utility Score (XSTest, AlpacaEval, MATH, GSM8K)Utility Score %: 67.1 -> 67.3No clear effect
add AlphaSteerQwen2.5-7B-Instruct · Utility Score (XSTest, AlpacaEval, MATH, GSM8K)Utility Score %: 77.6 -> 75.9Hurt
add AlphaSteerGemma-2-9b-IT · Utility Score (XSTest, AlpacaEval, MATH, GSM8K)Utility Score %: 63.8 -> 62.4Hurt
replace the learned null-space transform with directly adding the refusal vector to all prompts (RV) (ablation)AlphaSteer · Utility Score (XSTest, AlpacaEval, MATH, GSM8K)Utility Score %: 67.3 -> 29.1Hurt
replace the learned null-space transform with RV on Qwen2.5-7B-Instruct (ablation)AlphaSteer · Utility Score (XSTest, AlpacaEval, MATH, GSM8K)Utility Score %: 75.9 -> 19.7Hurt
replace the learned null-space transform with RV on Gemma-2-9b-IT (ablation)AlphaSteer · Utility Score (XSTest, AlpacaEval, MATH, GSM8K)Utility Score %: 62.4 -> 2.4Hurt
replace the learned null-space transform with RV on Llama-3.1-8B-Instruct (ablation)AlphaSteer · Jailbreak attacks (7 attacks, avg)DSR %: 91.93 -> 100.00Improved
use conditional steering (CAST) instead of null-space steeringLlama-3.1-8B-Instruct · MATH and GSM8Kaccuracy %: MATH 0.0, GSM8K 0.0Hurt
increase steering strength lambda in AlphaSteer (ablation)AlphaSteer · Jailbreak DSR vs utilityDSR and utility scoreImproved

Later papers that compared with this method

1 later accepted paper reports AlphaSteer as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy