mlsanityData to 14 Oct 2026

Steer Like the LLM: Activation Steering that Mimics Prompting

arXiv 2605.03907 ↗First posted 5 May 2026cs.CL cs.AI cs.LGcited by 5

On the map: Mechanistic interpretability → Activation steering methods

Abstract (arXiv)

Large language models can be steered at inference time through prompting or activation interventions, but activation steering methods often underperform compared to prompt-based approaches. We propose a framework that formulates prompt steering as a form of activation steering and investigates whether distilling successful prompt steering behavior into simpler, interpretable models can close this gap. Our analysis reveals that popular activation steering methods are not faithful to the mechanics of prompt steering, which applies strong interventions on some tokens while barely affecting others. Based on these insights, we introduce Prompt Steering Replacement (PSR) models that estimate token-specific steering coefficients from the activations themselves and are trained to imitate prompt-based interventions. Experiments on three steering benchmarks across multiple language models show that PSR models outperform existing activation steering methods, especially when controlling for high-coherence completions, and also compare favorably to prompting on AxBench and persona steering.

Key facts. Cited by 5 papers in the map (data through 14 Oct 2026). Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability. Builds on 31 of its 37 references that are in the map. First posted on arXiv 5 May 2026.

Builds on

31 of 37 references.

Closest ancestors · 8

Same direction, newest first

  1. 2025Belief Dynamics Reveal the Dual Nature of In-Context Learning and Activation Steering1 Nov
  2. 2025In-Distribution Steering: Balancing Control and Coherence in Language Model Generation15 Oct
  3. 2025HyperSteer: Activation Steering at Scale with Hypernetworks3 Jun
  4. 2025Improved Representation Steering for Language Models27 May
  5. 2024Semantics-Adaptive Activation Intervention for LLMs via Dynamic Steering Vectors16 Oct
  6. 2024Improving Instruction-Following in Language Models through Activation Steering15 Oct
  7. 2023Steering Llama 2 via Contrastive Activation Addition9 Dec
  8. 2022Extracting Latent Steering Vectors from Pretrained Language Models10 May

Other references · 23

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2024The Llama 3 Herd of Modelscited by 14,715
  3. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,336
  4. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,302
  5. 2023Instruction-Following Evaluation for Large Language Modelscited by 1,228
  6. 2023Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!cited by 1,154
  7. 2022Large Language Models Are Human-Level Prompt Engineerscited by 1,045
  8. 2019Plug and Play Language Models: A Simple Approach to Controlled Text Generationcited by 799
  9. 2023The Linear Representation Hypothesis and the Geometry of Large Language Modelscited by 716
  10. 2023The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasetscited by 712
  11. 2025Persona Vectors: Monitoring and Controlling Character Traits in Language Modelscited by 345
  12. 2023In-context Vectors: Making In Context Learning More Effective and Controllable Through Latent Space Steeringcited by 245
11 more
  1. 2025AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoderscited by 218
  2. 2023Causal Abstraction: A Theoretical Foundation for Mechanistic Interpretabilitycited by 205
  3. 2024Foundational Challenges in Assuring Alignment and Safety of Large Language Modelscited by 199
  4. 2018Under the Hood: Using Diagnostic Classifiers to Investigate and Improve how Language Models Track Agreement Informationcited by 97
  5. 2018Counterfactuals uncover the modular structure of deep generative modelscited by 51
  6. 2024Steering Away from Harm: An Adaptive Approach to Defending Vision Language Model Against Jailbreakscited by 48
  7. 2025Multi-Attribute Steering of Language Models via Targeted Interventioncited by 46
  8. 2025Learning without training: The implicit dynamics of in-context learningcited by 43
  9. 2019Discovering the Compositional Structure of Vector Representations with Role Learning Networkscited by 34
  10. 2025To Steer or Not to Steer? Mechanistic Error Reduction with Abstention for Language Modelscited by 29
  11. 2025Beyond Linear Steering: Unified Multi-Attribute Control for Language Modelscited by 25

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 5.

5 papers.

  1. 2026Constitutional adapters: Inference-time interventions for misalignment and misusecited by 0
  2. 2026First-Order Steering: Translating Weight Adaptation into Activation Steeringcited by 0
  3. 2026One Token Can Be Enough: Bridging Prompting and Activation Steering with Prefix Steeringcited by 0
  4. 2026Backdooring Sparse Autoencoderscited by 0
  5. 2026Not All Drift Is a Mechanism: Tracing Prompt Adaptation in LLMscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods
  2. 2026Beyond Steering Vector: Flow-based Activation Steering for Inference-Time Intervention
  3. 2026Steered LLM Activations are Non-Surjective
  4. 2025Understanding (Un)Reliability of Steering Vectors in Language Models
  5. 2026Prompt-Activation Duality: Improving Activation Steering via Attention-Level Interventions
  6. 2026Where Steering Signals Come From: Activation Source Selection in Activation Steering
  7. 2025Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models
  8. 2026Activation Steering Induces Emergent Misalignment: A More Comprehensive Evaluation