arXiv:2511. 07107v3 Announce Type: replace Abstract: Ensuring the safety of Large Language Models (LLMs) is critical for real-world deployment.
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:2511. 21214v4 Announce Type: replace-cross Abstract: Explicit safety policies can improve reasoning-model safety, but their effective coverage may lag behind evolving jailbreak strategies.
arXiv:2607. 23496v1 Announce Type: new Abstract: Safety-aligned large language models are trained to refuse harmful requests, yet embedding the same requests in particular scenarios can bypass their safeguards.
Safety-aligned large language models are trained to refuse harmful requests, yet embedding the same requests in particular scenarios can bypass their safeguards. Existing red-teaming methods empirically identify effective scenarios through observed attack outcomes, but why particular scenarios weaken refusal remains mechanistically unclear.
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.
The article surveys safety concerns for self‑evolving agents that continually update their internal state, such as model parameters and memories, from new interactions. It introduces the SAVER framework, which tracks reusable influence, adaptation, violations, exposure, and response to assess whether safety properties persist as agents evolve. The survey finds that legitimate state can become unsafe when its persistence, authority, or scope expands beyond its original conditions, and highlights gaps in current research on descendant repair and longitudinal evaluation.
arXiv:2607. 07903v1 Announce Type: cross Abstract: Large language models (LLMs) exhibit remarkable capabilities but remain highly vulnerable to adversarial prompts and jailbreak attacks.