paper-with-me

홈 › Papers

Furina: Fragmented Uncertainty-Driven Refusal Instability Attack

2026-05-24 · Tongxi Wu, Jian Zhang, Yang Gao arxiv

Safety alignment in large language models (LLMs) and multimodal large language models (MLLMs) is commonly assumed to operate as a near-binary threshold mechanism. We challenge this assumption by revealing that safety behavior is governed by an instability region where small perturbations induce stochastic refusal decisions rather than deterministic outcomes. We develop a multi-metric diagnostic framework combining external and internal signals to characterize this instability. Through systematic experiments, we identify a characteristic diagnostic signature: inputs in unstable regimes exhibit elevated output uncertainty yet decreased internal safety activation, a decoupling phenomenon that explains why detection-based defenses fail against sophisticated attacks. Building on this framework, we introduce Furina, a jailbreak attack that deliberately induces this signature through fragmented, scene-anchored prompts without model-specific optimization. Furina outperforms strong single-turn and multi-turn baselines on HarmBench and achieves competitive results on MM-SafetyBench, demonstrating that uncertainty amplification provides a principled and transferable mechanism for understanding safety vulnerabilities. Code is available at: https://github.com/0xCavaliers/Furina_Jailbreak.

📄 PDF Abstract BibTeX arXiv:2605.26158

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

FURINA: A Fully Customizable Role-Playing Benchmark via Scalable Multi-Agent Collaboration Pipeline

2025-10-08 · Haotian Wu, Shufan Jiang, Chios Chen, Yiyang Feng 외 arxiv

As large language models (LLMs) advance in role-playing (RP) tasks, existing benchmarks quickly become obsolete due to their narrow scope, outdated interaction paradigms, and limited adaptability across diverse applicati…

FURINA: Free from Unmergeable Router via LINear Aggregation of mixed experts

2025-09-18 · Jiayi Han, Liang Du, Yinda Chen, Xiao Kang 외 arxiv

The Mixture of Experts (MoE) paradigm has been successfully integrated into Low-Rank Adaptation (LoRA) for parameter-efficient fine-tuning (PEFT), delivering performance gains with minimal parameter overhead. However, a …

parameter-efficient fine-tuning

Do Retrieval Augmented Language Models Know When They Don't Know?

2025-09-01 · Youchao Zhou, Heyan Huang, Yicheng Liu, Rui Dai 외 arxiv

Existing large language models (LLMs) occasionally generate plausible yet factually incorrect responses, known as hallucinations. Two main approaches have been proposed to mitigate hallucinations: retrieval-augmented lan…

Breaking Safety at the Token Boundary: How BPE Tokenization Creates Exploitable Gaps in LLM Alignment

2026-05-01 · Tung-Ling Li, Hongliang Liu, Yuhao Wu arxiv

Character-level perturbations bypass safety alignment in modern LLMs despite leaving prompts human-readable. We identify and test a central structural mechanism: BPE tokenization fragments safety-critical words into sub-…

Ablating Safety: Mechanisms for Removing Alignment in Language Models for Security Applications

2026-05-17 · Isaac David, Arthur Gervais arxiv

Safety-aligned language models often refuse cybersecurity requests whose wording resembles misuse, even when the task is authorized and defensive. This makes security evaluation ambiguous: a failed answer may reflect mis…