arXiv AI September 7, 2026 By Yonghong Deng, Zhen Yang, Ping Jian, Xinyue Zhang, Zhongbin Guo, Chengzhi Li, Junxi Yin

The Struggle Between Continuation and Refusal: A Mechanistic Analysis of the Continuation-Triggered Jailbreak in LLMs

The paper investigates a continuation-triggered jailbreak in large language models, showing that moving an instruction suffix can markedly boost jailbreak success. By performing mechanistic interpretability at the attention‑head level, the authors reveal that the jailbreak arises from a competition between the model’s natural continuation drive and safety defenses learned during alignment. They introduce Head Competition Steering (HCS), an inference‑time technique that exploits this competition to suppress harmful outputs and distill the approach into a student model for efficient safety improvements.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

arXiv AI

Jul 2

HARC: Coupling Harmfulness and Refusal Directions for Robust Safety Alignment

arXiv:2607. 00572v1 Announce Type: new Abstract: Understanding how aligned LLMs internally represent safety is critical for diagnosing alignment vulnerabilities, as it explains why jailbreaks succeed and informs the design of robust alignment strategies.

By Shei Pern Chua, Fangzhao Wu

llms fine-tuning safety

arXiv AI

Aug 10

Minimal, Local, Causal Explanations for Jailbreak Success in Large Language Models

arXiv:2605. 00123v3 Announce Type: replace Abstract: Safety trained large language models (LLMs) can often be induced to answer harmful requests through jailbreak prompts.

By Shubham Kumar, Narendra Ahuja

llms benchmarks safety

arXiv AI

Jun 29

Robust Harmful Features Under Jailbreak Attacks: Mechanistic Evidence from Attention Head Specialization in Large Language Models

arXiv:2606. 28153v1 Announce Type: cross Abstract: Jailbreak attacks bypass LLM safety alignment, yet their mechanisms remain poorly understood.

By Yanchen Yin, Dongqi Han, Linghui Li

llms benchmarks safety

arXiv AI

Jul 1

The Geometry of Refusal: Linear Instability in Safety-Aligned LLMs

arXiv:2606. 22686v2 Announce Type: replace-cross Abstract: Modern Large Language Models (LLMs) rely on extensive safety alignment, yet the mechanistic basis of refusal remains opaque.

By Shivam Ratnakar, Kartikeya Vats

llms multimodal safety

arXiv AI

Jul 24

Incomplete Prompt Jailbreaks in Large Language Models

arXiv:2607. 20473v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly released as open-weight models with safeguards against harmful requests.

By Yeonjea Kim, Bumjin Park, Jaesik Choi

llms

arXiv AI

2d ago

AlcaTRAz - Anchored Tree-Rule Defense Against Jailbreaks

AlcaTRAz is a prompt‑level defense that uses rule trees to insert controlled character‑level perturbations into input text, disrupting jailbreak attacks without modifying or retraining the target LLM. It operates solely on the input, making it suitable for black‑box deployments, and was evaluated on 33 open‑weight models and 22 jailbreak types, outperforming three baseline defenses in 73.4 % of model‑attack combinations. While it significantly reduces high‑severity jailbreak success, it does not eliminate it and is intended as one layer of a broader defense strategy.

By Jakub Re\v{s}, Petr Ka\v{s}ka, Martin Pere\v{s}\'ini, Martin Ukrop, Kamil Malinka

llms benchmarks safety

One email a morning, machine-written

Related stories

HARC: Coupling Harmfulness and Refusal Directions for Robust Safety Alignment

Minimal, Local, Causal Explanations for Jailbreak Success in Large Language Models

Robust Harmful Features Under Jailbreak Attacks: Mechanistic Evidence from Attention Head Specialization in Large Language Models

The Geometry of Refusal: Linear Instability in Safety-Aligned LLMs

Incomplete Prompt Jailbreaks in Large Language Models

AlcaTRAz - Anchored Tree-Rule Defense Against Jailbreaks