EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Risk-Manifold Analytics for Detecting Deceptive Reasoning in Edge-Deployed Large Language Models

Xiangyuan Li, Lijuan Zhang, Shitong Ye

Journal of AI Analytics and Applications · 30.12.2025 · 37-53

DOI: 10.63646/ofqk9310 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of AI Analytics and Applications, online erschienen am 30.12.2025. Verfasst von Xiangyuan Li und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The proliferation of compact language models on resource-constrained edge hardware has created urgent demand for safety monitoring architectures that operate entirely offline, without reliance on cloud-based adjudication or heavyweight teacher models. Existing deceptive alignment detection methods reduce the problem to binary classification over Chain-of-Thought reasoning traces, a formulation that ignores the continuous nature of deceptive reasoning and requires external oracle annotation. This paper introduces Risk-Manifold Analytics (RMA), a geometric framework that characterises deceptive reasoning as a structured topological risk space rather than a discrete class boundary. RMA employs a three-stage pipeline: entropy-filtered autonomous label generation, manifold-constrained supervised fine-tuning with Triplet Loss optimisation, and frozen-monitor constrained proximal policy optimisation. A lightweight risk projector (0.1% of backbone parameters) maps Chain-of-Thought hidden states onto a 128-dimensional unit hypersphere where deceptive and safe reasoning clusters are geometrically separable, enabling multi-dimensional risk scoring that captures gradual deceptive transitions from surface hedging to objective substitution. Evaluated on DeceptionBench across five deception taxonomies with 180 adversarial scenarios, RMA achieves a Deception Tendency Rate (DTR) of 36.96% on Gemma-3-4B-IT under full offline operation on NVIDIA Jetson Orin Nano hardware consuming only 7.5 W active power. Ablation studies confirm a 2.33 percentage point improvement over binary cross-entropy baselines, while cross-model validation across five architectures spanning 2B to 7B parameters demonstrates consistent DTR reductions of 3.74–4.44 percentage points. The proposed framework establishes a theoretically grounded geometric foundation for autonomous, privacy-preserving deceptive alignment monitoring suitable for deployment in safety-critical edge environments.

Zitieren

APA

Li, X., Zhang, L., & Ye, S. (2025). Risk-Manifold Analytics for Detecting Deceptive Reasoning in Edge-Deployed Large Language Models. Journal of AI Analytics and Applications, 37-53. https://doi.org/10.63646/ofqk9310

Vancouver

Li X, Zhang L, Ye S. Risk-Manifold Analytics for Detecting Deceptive Reasoning in Edge-Deployed Large Language Models. Journal of AI Analytics and Applications. 2025:37-53. doi:10.63646/ofqk9310

BibTeX
@article{Li2025,
  author = {Li, Xiangyuan and Zhang, Lijuan and Ye, Shitong},
  title = {Risk-Manifold Analytics for Detecting Deceptive Reasoning in Edge-Deployed Large Language Models},
  journal = {Journal of AI Analytics and Applications},
  year = {2025},
  pages = {37--53},
  doi = {10.63646/ofqk9310}
}
Kostenlos starten →