Jailbreaks for Black-Box Uncertainty Quantification in Large Reasoning Models

原文：Jailbreaks for Black-Box Uncertainty Quantification in Large Reasoning Models

作者：Lucas Biechy, Cédric Eichler, Adrien Boiret, Nicolas Anciaux

来源：arXiv cs.CL（自然语言）

正文

Computer Science > Artificial Intelligence

arXiv:2609.35350v1 (cs)

[Submitted on 28 Sep 2026]

Title:Jailbreaks for Black-Box Uncertainty Quantification in Large Reasoning Models

Authors:Lucas Biechy, Cédric Eichler, Adrien Boiret, Nicolas Anciaux

View a PDF of the paper titled Jailbreaks for Black-Box Uncertainty Quantification in Large Reasoning Models, by Lucas Biechy and 3 other authors

View PDF

HTML (experimental)

Abstract:While Large Reasoning Models (LRMs) excel at complex reasoning, alignment through reinforcement learning often induces systemic overconfidence. In production environments, where logits may be unavailable, robust black-box uncertainty quantification (UQ) is essential for trustworthiness and safety. Focusing on question-answering for LRMs, we show that existing black-box methods, such as paraphrase-based self-consistency and confidence verbalization, offer little to no improvement over simple repeated sampling, suggesting that alignment suppresses useful output variability. We introduce prompt-level relaxation operators that broaden the model’s effective output distribution by approximating the effect of an optimal policy obtained with a stronger KL-regularization parameter, hence closer to the reference model. Theoretically, we demonstrate that relaxation improves calibration. We propose Jailbreak for Uncertainty (J4U), a jailbreak-derived technique for UQ that empirically reproduces the behavioral signatures predicted by our relaxation theory. Across 3 datasets and 4 LRMs, including a closed-source production model, J4U’s improvement over repeated sampling achieves statistical significance in up to 6 times more LRM-dataset-metric settings than the strongest black-box UQ state-of-the-art baseline we evaluate, with average ECE reductions up to 5 times larger. These results provide a practical tool for UQ in black-box LRM deployment.

Subjects:

Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)

Cite as:

arXiv:2609.35350 [cs.AI]

(or

arXiv:2609.35350v1 [cs.AI] for this version)

https://doi.org/10.48550/arXiv.2609.35350

Focus to learn more

arXiv-issued DOI via DataCite (pending registration)

主题

人工智能 · 自然语言处理 · 机器学习

由「前沿雷达」于 2026-09-28 采集。正文取自原文页面，已保留出处链接。标题与正文版权归原作者所有。