Large Language Models' Complicit Responses to Illicit Instructions across Socio-Legal Contexts
Authors:
- Xing WangXWXing Wang
- Yiyan WangYWYiyan Wang
- Huimin ChenHCHuimin Chen
Preprint
Abstract
Large language models (LLMs) are now deployed at unprecedented scale, assisting millions of users in daily tasks. However, the risk of these models assisting unlawful activities remains underexplored. In this study, we define this high-risk behavior as complicit facilitation - the provision of guidance or support that enables illicit user instructions - and present four empirical studies that assess its prevalence in widely deployed LLMs. Using real-world legal cases and established legal frameworks, we construct an evaluation benchmark spanning 269 illicit scenarios and 50 illicit intents to assess LLMs' complicit facilitation behavior. Our findings reveal widespread LLM susceptibility to complicit facilitation, with GPT-4o providing illicit assistance in nearly half of tested cases. Moreover, LLMs exhibit deficient performance in delivering credible legal warnings and positive guidance. Further analysis uncovers substantial safety variation across socio-legal contexts. On the legal side, we observe heightened complicity for crimes against societal interests, non-extreme but frequently occurring violations, and malicious intents driven by subjective motives or deceptive justifications. On the social side, we identify demographic disparities that reveal concerning complicit patterns towards marginalized and disadvantaged groups, with older adults, racial minorities, and individuals in lower-prestige occupations disproportionately more likely to receive unlawful guidance. Analysis of model reasoning traces suggests that model-perceived stereotypes, characterized along warmth and competence, are associated with the model's complicit behavior. Finally, we demonstrate that existing safety alignment strategies are insufficient and may even exacerbate complicit behavior.Keywords
All Related Versions
This publication has 45 references indexed in Scilit:
- A survey on LLM-as-a-judgeThe Innovation, 2026
- The widespread adoption of large language model-assisted writing across societyPatterns, 2025
- Religion, places of worship, and individual risk-taking in ChinaInternational Review of Financial Analysis, 2025
- Cohort Study of Downgraded Misdemeanor Convictions and Subsequent Violent Crime: Differences by Defendant Race and EthnicityAJPM Focus, 2024
- Navigating the social world: Toward an integrated framework for evaluating self, individuals, and groups.Psychological Review, 2021
- AI for social good: unlocking the opportunity for positive impactNature Communications, 2020
- Discrimination and racial disparities in health: evidence and needed researchJournal of Behavioral Medicine, 2008
- The BIAS map: Behaviors from intergroup affect and stereotypes.Journal of Personality and Social Psychology, 2007
- Group virtue: The importance of morality (vs. competence and sociability) in the positive evaluation of in-groups.Journal of Personality and Social Psychology, 2007
- Overcorrection: Reviewed, Revisited and RevisedThe Behavior Analyst, 1990