The paper introduces MINT‑Safe, a new open‑source dataset of 11,270 multi‑image dialogues and 500 refusal VQA pairs designed to expose safety risks in multi‑modal large language models during open‑ended conversations. It also proposes TAD‑Align, a turn‑aware dual‑objective reward framework that dynamically up‑weights dialogue turns with inconsistent safety behavior, improving safety metrics on Qwen2.5‑VL‑7B‑Instruct and LLaVA‑Next‑7B. The results show over 10% reduction in attack success rate and notable gains in harmlessness and helpfulness while maintaining overall model performance.
By Han Zhu, Jiale Chen, Chengkun Cai, Shengjie Sun, Haoran Li, Yujin Zhou, Chi-Min Chan, Pengcheng Wen, Lei Li, Yike Guo, Sirui Han
arXiv:2606. 08531v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly evolving from simple text-based interaction systems into LLM agents that can maintain memory, use tools, access external environments, and execute tasks.
By Lu Jia, Haibo Tong, Feifei Zhao, Jindong Li, Dongqi Liang, Ping Wu, Qian Zhang, Yi Zeng
arXiv:2609.35812v1 Announce Type: new
Abstract: In-car conversational assistants (ICAs) are increasingly integrated into vehicles to support route planning, vehicle control, and information access. E...
By Vaishnav Negi, Lev Sorokin, Soroosh Tayebi Arasteh, Andrea Stocco
arXiv:2606. 08531v2 Announce Type: replace Abstract: Large language models (LLMs) are increasingly evolving from simple text-based interaction systems into LLM agents that can maintain memory, use tools, access external environments, and execute tasks.
By Lu Jia, Haibo Tong, Feifei Zhao, Jindong Li, Dongqi Liang, Ping Wu, Qian Zhang, Yi Zeng
arXiv:2604. 17301v2 Announce Type: replace-cross Abstract: Detecting harmful content in multi turn dialogue requires reasoning over the full conversational context rather than isolated utterances.
By Juhyeon Lee, Wonduk Seo, Junseo Koh, Seunghyun Lee, Haihua Chen, Yi Bu
arXiv:2607. 01153v1 Announce Type: cross Abstract: Safety evaluations for language models increasingly depend on judgments about ambiguous natural-language behaviour: whether a model has followed an instruction, refused appropriately, complied with a policy, resisted an embedded command, or misreported progress in an agentic task.
By Brett Reynolds
arXiv:2603. 15684v2 Announce Type: replace-cross Abstract: Safety alignment in large language models is typically evaluated under isolated queries, yet real-world use is inherently multi-turn.
By Pengcheng Li, Jie Zhang, Tianwei Zhang, Han Qiu, Zhang kejun, Weiming Zhang, Nenghai Yu, Wenbo Zhou
arXiv:2606. 10380v1 Announce Type: cross Abstract: Real-world crisis intervention is inherently conversational, yet existing research largely focuses on static texts.
By Grace Byun, Abigail Lott, Rebecca Lipschutz, Sean T. Minton, Elizabeth A. Stinson, Jinho D. Choi
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
arXiv:2607. 22635v1 Announce Type: new Abstract: Target-oriented dialogue systems have demonstrated strong capabilities in completing user goals through interactive conversations.
By Xuzhao Geng, Haozhao Wang, Xuelian Li, Zhenyu Yang, Haonan Lu, Rui Zhang, Ruixuan Li
Safeguarding language model agents requires assessing complete execution trajectories under context-dependent safety policies. Existing policy-aware safeguards mainly rely on prompting or supervised f...
RePolicy is a reinforcement learning approach designed to invoke safety policies for language model agents by evaluating entire execution trajectories within context-dependent policy libraries. It generates policy-grounded rationales and safety judgments, and is initialized with the PolicyTraj-20K dataset before fine-tuning via GRPO with verifiable rewards and policy-context perturbation. Experiments on six safety benchmarks demonstrate strong safety-detection performance and robust policy invocation across varying contexts.
By Houcheng Jiang, Boxuan Zhang, Qiyong Zhong, Junfeng Fang, Xiang Wang, Xiangnan He