The paper identifies a vulnerability in large language models where harmful intent can be hidden within benign narratives, a phenomenon termed Semantic Camouflage. By examining latent activation patterns across several small language model families, the authors discover an "Intent Horizon"—a layer depth where harmful intent representations collapse. They propose Latent Intent Verification (LIV), a lightweight probing defense that detects harmful intent in early layers and outperforms existing guardrails on the PKU-SafeRLHF dataset.
By Md. Hasib Ur Rahman
arXiv:2603. 19423v2 Announce Type: replace-cross Abstract: Large language model (LLM) agents increasingly rely on external tools (file operations, API calls, database transactions) to autonomously complete complex multi-step tasks.
By Shawn Li, Yue Zhao
arXiv:2603. 00801v2 Announce Type: replace Abstract: Language agents increasingly act as web-enabled systems that search, browse, and synthesize information from diverse sources.
By Shrey Shah, Levent Ozgur
arXiv:2609.38270v1 Announce Type: cross
Abstract: Advancing beyond traditional static scoring models, LLM-powered agentic recommender systems (LLM-ARS) instantiate users and items as autonomous agent...
By Yurong Hao, Wen Zhou, Guowei Guan, Tiantong Wu, Fuyao Zhang, Wei Yang Bryan Lim
The paper introduces FAB, an attack that uses meta‑learning to embed dormant adversarial behaviors into large language models (LLMs). These behaviors remain inactive until the model is finetuned by downstream users, at which point the model can exhibit unwanted actions such as unsolicited advertising, jailbreakability, or over‑refusal. FAB is shown to be effective across multiple LLMs and resilient to various finetuning settings.
By Thibaud Gloaguen, Mark Vero, Robin Staab, Martin Vechev
arXiv:2606. 29657v1 Announce Type: new Abstract: As AI systems become more capable, training procedures that optimize for downstream outcomes risk introducing implicit agency: goal-directed behavior that designers never specified.
By Yoshua Bengio, Oliver Richardson, Tom\'a\v{s} Gaven\v{c}iak, Michael Cohen, Rory Svarc, Damiano Fornasiere, Gael Gendron, David Hyland, Aton Kamanda, Adam Oberman, Francis Rhys Ward, Anna Gaven\v{c}iak, Jacob Livingston Slosser, Vincent Mai, Iulian Serban, Joumana Ghosn
arXiv:2608. 00566v1 Announce Type: new Abstract: Post-hoc model explainers such as LIME, SHAP, and Integrated Gradients are widely deployed to audit models in high-stakes sensitive domains, including finance, healthcare, and social welfare.
By Niraj Kumar, Harsh Kasyap
arXiv:2604. 08169v2 Announce Type: replace Abstract: Alignment in LLMs is more brittle than commonly assumed: misalignment can be induced by adversarial prompts, benign fine-tuning, emergent misalignment, and goal misgeneralization.
By Niklas Herbster, Martin Zborowski, Alberto Tosato, Gauthier Gidel, Tommaso Tosato
As AI systems become more capable, training procedures that optimize for downstream outcomes risk introducing implicit agency: goal-directed behavior that designers never specified. We present a formal safety argument for the Scientist AI (SAI) Predictor, trained to approximate the Bayesian posterior conditioned on a dataset of "epistemically contextualized" natural-language statements.
The paper introduces CS‑VAR, a Cross‑Session Evidence‑Aware Retrieval‑Augmented Detector that uses a lightweight, domain‑specific model for fast session‑level risk inference in live streaming. The model is trained with guidance from a Large Language Model that reasons over retrieved cross‑session behavioral evidence, enabling the small model to recognize recurring risk patterns and perform structured risk assessment. Experiments on large industrial datasets and online validation show that CS‑VAR achieves state‑of‑the‑art performance while providing interpretable, localized signals for real‑world moderation.
By Yiran Qiao, Xiang Ao, Jing Chen, Yang Liu, Qiwei Zhong, Qing He
arXiv:2607. 26820v1 Announce Type: new Abstract: As large language models (LLMs) evolve from standalone assistants into autonomous agents, ensuring their safety requires shifting beyond pointwise risk assessment to understand how risks emerge and unfold over long-horizon trajectories.
By Shi Lin, Peng Qian, Dinghao Liu, Renjie Sun, Sifan Wu, Dezhang Kong, Chenpei Wang, Xun Wang
The paper introduces a new evaluation setting called scenario‑level out‑of‑distribution (SL‑OOD) detection for SMS and voice phishing, where entire attack scenarios are omitted from training while the label space stays fixed. It shows that high in‑distribution performance does not guarantee robustness to unseen scenarios, attributing this to scenario memorization. The authors propose ECoG, an evidence‑consistent generative framework that uses evidence‑span supervision and a rationale‑label consistency objective, achieving notable improvements in Macro‑F1, reduced prediction‑rationale inconsistency, and higher token‑level overlap with reference evidence.
By San Kim, JinYeong Bak