Prompt-Induced Bias in LLM Security Recommendations: A Quantitative Behavioral Study
Computer Science Bulletin · 28.02.2026 · 9 (2) · 277-283
DOI: 10.71465/csb222 ↗Kurz gefasst
Fachartikel in der Zeitschrift Computer Science Bulletin, online erschienen am 28.02.2026. Verfasst von Liam Hartley und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
This study explores how prompt phrasing influences the securityrecommendations produced by large language models. Using 4,200controlled prompt variations,we analyze how prescriptive wording, threat emphasis, risk framing, and confidence cues shape LLM suggestions related to input validation, authentication, and cryptographic configurations. Results reveal substantial prompt-induced bias: risk-framing prompts increase conservative recommendations by 31%, while permissive wording reduces strict security advice by 23%. Moreover, confidence-leading prompts cause over-assertive but incorrect recommendations in 12% of cases. A mitigation framework based on neutralization templates reduces bias variance by 64%. These findings highlight the importance of prompt neutrality for dependable security decision support.
Zitieren
APA
Hartley, L., Zhao, E., & McMillan, S. (2026). Prompt-Induced Bias in LLM Security Recommendations: A Quantitative Behavioral Study. Computer Science Bulletin, 9(2), 277-283. https://doi.org/10.71465/csb222
Vancouver
Hartley L, Zhao E, McMillan S. Prompt-Induced Bias in LLM Security Recommendations: A Quantitative Behavioral Study. Computer Science Bulletin. 2026;9(2):277-283. doi:10.71465/csb222
BibTeX
@article{Hartley2026,
author = {Hartley, Liam and Zhao, Emily and McMillan, Sophie},
title = {Prompt-Induced Bias in LLM Security Recommendations: A Quantitative Behavioral Study},
journal = {Computer Science Bulletin},
year = {2026},
volume = {9},
number = {2},
pages = {277--283},
doi = {10.71465/csb222}
}
EUVIMED