# FaithSteer-BENCH: A Deployment-Aligned Stress-Testing Benchmark for Inference-Time Steering

arXiv: https://arxiv.org/abs/2603.18329 · HTML: https://mlsanity.com/paper/2603.18329/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 13 of its 23 references that are in the map.
- First posted on arXiv 18 Mar 2026.

## Builds on (most cited first, 13 of 13)

- [HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal](https://mlsanity.com/paper/2402.04249/) (2024)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Steering Language Models With Activation Engineering](https://mlsanity.com/paper/2308.10248/) (2023)
- [Discovering Language Model Behaviors with Model-Written Evaluations](https://mlsanity.com/paper/2212.09251/) (2022)
- [AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders](https://mlsanity.com/paper/2501.17148/) (2025)
- [Steering Language Model Refusal with Sparse Autoencoders](https://mlsanity.com/paper/2411.11296/) (2024)
- [A Unified Understanding and Evaluation of Steering Methods](https://mlsanity.com/paper/2502.02716/) (2025)
- [Activation Scaling for Steering and Interpreting Language Models](https://mlsanity.com/paper/2410.04962/) (2024)
- [Evaluating the Prompt Steerability of Large Language Models](https://mlsanity.com/paper/2411.12405/) (2024)
- [CLaS-Bench: A Cross-Lingual Alignment and Steering Benchmark](https://mlsanity.com/paper/2601.08331/) (2026)
- [Efficient and accurate steering of Large Language Models through attention-guided feature learning](https://mlsanity.com/paper/2602.00333/) (2026)
- [COLD-Steer: Steering Large Language Models via In-Context One-step Learning Dynamics](https://mlsanity.com/paper/2603.06495/) (2026)

## Built on by (most cited first, 2 of 2)

- [Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods](https://mlsanity.com/paper/2610.07722/) (2026)
- [MAxBench: A Multinomial Concept Recovery Benchmark](https://mlsanity.com/paper/2609.13072/) (2026)

## Abstract (arXiv)

Inference-time steering is widely regarded as a lightweight and parameter-free mechanism for controlling large language model (LLM) behavior, and prior work has often suggested that simple activation-level interventions can reliably induce targeted behavioral changes. However, such conclusions are typically drawn under relatively relaxed evaluation settings that overlook deployment constraints, capability trade-offs, and real-world robustness. We therefore introduce FaithSteer-BENCH, a stress-testing benchmark that evaluates steering methods at a fixed deployment-style operating point through three gate-wise criteria: controllability, utility preservation, and robustness. Across multiple models and representative steering approaches, we uncover several systematic failure modes that are largely obscured under standard evaluation, including illusory controllability, measurable cognitive tax on unrelated capabilities, and substantial brittleness under mild instruction-level perturbations, role prompts, encoding transformations, and data scarcity. Gate-wise benchmark results show that existing methods do not necessarily provide reliable controllability in deployment-oriented practical settings. In addition, mechanism-level diagnostics indicate that many steering methods induce prompt-conditional alignment rather than stable latent directional shifts, further explaining their fragility under stress. FaithSteer-BENCH therefore provides a unified benchmark and a clearer analytical lens for future method design, reliability evaluation, and deployment-oriented research in steering.