Spurious Prompts: Can Irrelevant Prompts Steer Large Language Models?

2026-05-28 · Pawel Batorski, Abtin Pourhadi, Jerzy Sarosiek, Przemyslaw Spurek, Paul Swoboda arxiv

Large language models are highly sensitive to prompts, but this sensitivity is usually studied through task-relevant instructions, demonstrations, or reasoning cues. In this paper, we study a different form of prompt sensitivity: whether prompts that are semantically unrelated to the task can nevertheless steer model behavior. We call them spurious prompts and show their surprising efficacy. We also propose a simple black-box search procedure for discovering them. Across reasoning and question-answering benchmarks, using models ranging from 0.8B to 27B parameters and spanning three model families, we show that spurious prompts can improve performance, often matching or outperforming standard prompting baselines and task-aware prompt optimization. We further show that they can steer models toward unintended behaviors, such as repeatedly selecting the first answer option, producing incorrect answers, returning an even, prime or small number without explicitly instructing the model to do so. These findings reveal a new kind of prompt sensitivity: LLMs can be systematically steered by prompts that are unrelated to the task they are asked to solve. Our code is available at https://github.com/Batorskq/spurious

📄 PDF Abstract BibTeX arXiv:2605.29678

Code (0)

등록된 구현이 없습니다.

Prompt Perturbation in Retrieval-Augmented Generation based Large Language Models

2024-02-11 · Zhibo Hu, Chen Wang, Yanfeng Shu, Helen 외

The robustness of large language models (LLMs) becomes increasingly important as their use rapidly grows in a wide range of domains. Retrieval-Augmented Generation (RAG) is considered as a means to improve the trustworth…

RAG Retrieval Retrieval-augmented Generation Text Generation

Feature-Guided SAE Steering for Refusal-Rate Control using Contrasting Prompts

2025-10-26 · Samaksh Bhargav, Zining Zhu arxiv

Large Language Model (LLM) deployment requires guiding the LLM to recognize and not answer unsafe prompts while complying with safe prompts. Previous methods for achieving this require adjusting model weights along with …

HyperSteer: Activation Steering at Scale with Hypernetworks

2025-06-03 · Jiuding Sun, Sidharth Baskaran, Zhengxuan Wu, Michael Sklar 외

Steering language models (LMs) by modifying internal activations is a popular approach for controlling text generation. Unsupervised dictionary learning methods, e.g., sparse autoencoders, can be scaled to produce many s…

Dictionary Learning Text Generation

Towards Single-Source Domain Generalized Object Detection via Causal Visual Prompts

2025-10-22 · Chen Li, Huiying Xu, Changxin Gao, Zeyu Wang 외 arxiv

Single-source Domain Generalized Object Detection (SDGOD), as a cutting-edge research topic in computer vision, aims to enhance model generalization capability in unseen target domains through single-source domain traini…

Domain Generalization Data Augmentation Domain Adaptation Object Detection

CASP: Few-Shot Class-Incremental Learning with CLS Token Attention Steering Prompts

2026-01-23 · Shuai Huang, Xuhan Lin, Yuwu Lu arxiv

Few-shot class-incremental learning (FSCIL) presents a core challenge in continual learning, requiring models to rapidly adapt to new classes with very limited samples while mitigating catastrophic forgetting. Recent pro…

Few-Shot Class-Incremental Learning Continual Learning