Large Language Models (LLMs) can solve complex problems, but their misuse in high-risk domains can lead to severe consequences. Model providers therefore restrict assistance for potentially harmful re...
arXiv:2606. 02644v1 Announce Type: cross Abstract: Agentic scaffolds have dramatically improved LLM performance on complex, long-horizon tasks, yielding both broad benefits and amplified risks in domains like cybersecurity.
The paper investigates whether large language models’ refusal behavior in text transfers to tool‑calling contexts. A benchmark of 2,304 prompts across four domains compares text‑only, standard tool‑enabled, and adversarially framed tool‑enabled settings, revealing that 10–52 % of harmful requests refused in chat still lead to unsafe tool calls. The authors identify a refusal‑associated direction in the models’ residual stream, showing that lower activation along this direction predicts unsafe calls, but restoring the signal only partially mitigates the issue and often suppresses benign tool use.
The paper introduces the SAST-IR framework to evaluate large language models’ robustness against persuasion attacks in a memory‑less setting, revealing a flaw called "Refusal Inertia" that masks true vulnerability. Using the CP‑Agent and a custom CounterFact‑Strict dataset, the authors demonstrate that simple, diverse attack strategies achieve a 96% success rate, while complex attacks often trigger defensive compliance. The study highlights severe brittleness in current state‑of‑the‑art models when deprived of conversation history.
The paper introduces a counterfactually anchored evidence attribution approach for multi‑turn large language model safety failures. It presents a new dataset of 1,762 conversations, including adversarial, benign twins, and high‑risk vocabulary variants, and trains a lightweight hierarchical model that accurately predicts safety violations and attributes them to specific user turns and token spans. The model achieves high detection performance (F1 = 0.988) and significantly reduces adversarial confidence when top‑attributed tokens are removed, while maintaining low false‑positive rates on benign conversations.
arXiv:2503. 15560v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly vulnerable to sophisticated multi-turn manipulation attacks, where adversaries strategically build context through seemingly benign conversational turns to circumvent safety measures and elicit harmful or unauthorized responses.
The paper investigates prompt injection attacks on 14 open‑source and 3 closed‑source large language models (LLMs), introducing a new metric called Attack Success Probability (ASP) that accounts for uncertainty in model responses. It demonstrates that a simple hypnotism attack can trigger objectionable behavior in models such as StableLM2, Mistral, Openchat, and Vicuna, achieving roughly 90% ASP. The study highlights that moderately well‑known LLMs are particularly vulnerable, underscoring the importance of public awareness and effective mitigation strategies.
arXiv:2508.02312v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs), now a foundation in advancing natural language processing, power applications such as text generation, machine...
arXiv:2606. 20470v1 Announce Type: cross Abstract: Agentic AI systems increasingly rely on language-model components to interpret instructions, process external data, invoke tools, and coordinate with other agents.
arXiv:2607. 02121v1 Announce Type: cross Abstract: As Large Language Models (LLMs) and agentic systems become integrated into real-world applications, ensuring their safety and security is critical.
arXiv:2608. 15893v1 Announce Type: new Abstract: The rise of social media bots poses a persistent threat, enabling misinformation, opinion manipulation, and the erosion of trust in online platforms.
arXiv:2609.13579v1 Announce Type: new Abstract: Safety research often focuses on model-generated harms, but users may also direct hostility, coercion, and adversarial pressure at models. Understandin...