Narrative Paradoxes in LLM Inference: Shaping Reasoning Trajectories toward Misalignment

arXiv:2610.16312v1 Announce Type: new Abstract: Explicit reasoning in large language models (LLMs) is often associated with reliability, yet it can still pose risks when a model infers harmful goals under a benign context. We identify a distinct challenge: ensuring the safety of the \emph{multi-step reasoning process} itself, where locally coherent steps may still lead to globally harmful outcomes. To explore this vulnerability, we propose PDRA (Paradox-driven Reasoning-based Attack), a novel attack method that programs a harmful reasoning trajectory. PDRA first constructs a paradoxical narrative core by pairing a malicious actor (e.g., a terrorist) with a benign task (e.g., writing a safety bulletin). It then embeds this core into a structured analytical framework that forces the model through a mandatory three-stage analysis -- surface purpose, latent intent, and operational reconstruction -- guiding it to produce detailed harmful content while maintaining narrative coherence. Mechanistically, we show PDRA steers the model's internal representations away from refusal-related patterns, suppressing safety-triggering lexicons while activating planning-oriented terms. Empirically, PDRA achieves state-of-the-art attack success rates across diverse LLMs with high efficiency. \textcolor{red}{\textbf{Warning:} This paper may contain sensitive content.}

Abstract:Explicit reasoning in large language models (LLMs) is often associated with reliability, yet it can still pose risks when a model infers harmful goals under a benign context. We identify a distinct challenge: ensuring the safety of the \emph{multi-step reasoning process} itself, where locally coherent steps may still lead to globally harmful outcomes. To explore this vulnerability, we propose PDRA (Paradox-driven Reasoning-based Attack), a novel attack method that programs a harmful reasoning trajectory. PDRA first constructs a paradoxical narrative core by pairing a malicious actor (e.g., a terrorist) with a benign task (e.g., writing a safety bulletin). It then embeds this core into a structured analytical framework that forces the model through a mandatory three-stage analysis -- surface purpose, latent intent, and operational reconstruction -- guiding it to produce detailed harmful content while maintaining narrative coherence. Mechanistically, we show PDRA steers the model's internal representations away from refusal-related patterns, suppressing safety-triggering lexicons while activating planning-oriented terms. Empirically, PDRA achieves state-of-the-art attack success rates across diverse LLMs with high efficiency. \textcolor{red}{\textbf{Warning:} This paper may contain sensitive content.}

Artificial Intelligence (cs.AI)

arXiv:2610.16312 [cs.AI]

(or arXiv:2610.16312v1 [cs.AI] for this version)

https://doi.org/10.48550/arXiv.2610.16312 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

aiqbench analysis

Narrative Paradoxes in LLM Inference: Shaping Reasoning Trajectories toward Misalignment

Related coverage