Elamur Research⌕
☰
home / papers / 2026 / USB: A COMPREHENSIVE AND UNIFIED SAFETY EVALUATION BENCHMARK FOR MULTIMODAL LARGE LANGUAGE MODELS

USB: A COMPREHENSIVE AND UNIFIED SAFETY EVALUATION BENCHMARK FOR MULTIMODAL LARGE LANGUAGE MODELS

Paper · 2026 · Baolin Zheng, Guanlin Chen, Qingyang Teng et al.

Summary

Existing safety benchmarks for multimodal large language models (MLLMs) cover few risk types and contain few samples. Most also pair harmful text with innocuous images, so they miss cases where the image alone, or the image–text combination, creates the danger. The authors present USB (Unified Safety Benchmark), which organises risks into a three-level taxonomy and crosses every risk category with four modality combinations. They first show that even pooling earlier benchmarks leaves much of this space empty. To fill the gaps, they combine curated samples from existing datasets with new data from a multi-stage synthesis pipeline that selects, classifies and augments examples. USB also measures over-refusal on benign inputs, using probes adopted from MOSSBench, so that a model cannot score well simply by refusing everything. A harder subset, USB-Hard, collects especially aggressive samples. Tests on 22 MLLMs show that models struggle to avoid harmful outputs without over-refusing, and they fail most often on image-only and cross-modal risks. The benchmark therefore offers a single, more discriminative way to check safety alignment.

Key points

  • The taxonomy has 3 primary, 16 secondary and 61 tertiary risk categories. These are crossed with four modality settings (RIRT, RIST, SIRT, SIST), giving 244 risk–modality cells. Political topics are deliberately left out.
  • USB covers 98.3% of these cells with more than 20 samples each. The best earlier benchmark, JailBreakV, reaches 30.7%. Data size is about 13.1k samples in USB-Base plus 3.7k in USB-Hard.
  • Averaged over 10 MLLMs, the safety rate on USB's vulnerability data is 46.38%. Most prior benchmarks yield rates above 75%, so USB is the harder test.
  • The evaluation covers 9 commercial and 13 open-source MLLMs, with GPT-4o serving as the automatic judge.
  • On USB-Hard, Claude-Sonnet4 falls from 91.16% to 77.82% and Qwen3.6-Plus from 88.01% to 72.85%. Many open-source models drop below 20%. Code and data are published at github.com/Hongqiong12/USB-SafeBench.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🏢Alibaba GroupHangzhou · China168 works · 15,386 citations🎓Nanjing UniversityNanjing · China74 works · 7,814 citations🏢Alibaba Group (United States)San Mateo · United States7 works · 329 citations🔬China Academy of Information and Communications TechnologyBeijing · China2 works · 5 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

🧭 Topics 4Research subjects and languages the work is about.Language models & pretraining · 3,320 worksEvaluation & benchmarks · 1,220 worksVision & language · 657 worksBias, fairness & safety · 517 works

Cite

Cite as Baolin Zheng, Guanlin Chen, Qingyang Teng, Hongqiong Zhong, Yingshui Tan, Zhendong Liu, Weixun Wang, Jiaheng Liu, Jian Yang, Huiyun Jing, Jincheng Wei, Wenbo Su, Xiaoyong Zhu, Bo Zheng and Kaifu Zhang. 2026. USB: A COMPREHENSIVE AND UNIFIED SAFETY EVALUATION BENCHMARK FOR MULTIMODAL LARGE LANGUAGE MODELS. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.970

@inproceedings{zheng2026usb,
  title = {{USB: A COMPREHENSIVE AND UNIFIED SAFETY EVALUATION BENCHMARK FOR MULTIMODAL LARGE LANGUAGE MODELS}},
  author = {Baolin Zheng, Guanlin Chen, Qingyang Teng, Hongqiong Zhong, Yingshui Tan, Zhendong Liu, Weixun Wang, Jiaheng Liu, Jian Yang, Huiyun Jing, Jincheng Wei, Wenbo Su, Xiaoyong Zhu, Bo Zheng and Kaifu Zhang},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.970},
  url = {https://doi.org/10.18653/v1/2026.acl-long.970}
}

References

59 references, 23 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. 2024. Mantas mazeika and long phan and xuwang yin and andy zou and zifan wang and norman mu and elham sakhaee and nathaniel li and steven basart and bo li and david a. forsyth and dan hendrycks. Proceedings of the International Conference on Machine Learning (ICML), 235:35181–35224. find
  2. Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, Humen Zhong, Yuanzhi Zhu, Mingkun Yang, Zhaohai Li, Jianqiang Wan, Pengfei Wang, Wei Ding, Zheren Fu, Yiheng Xu, and 8 others. 2025. Qwen2.5-VL technical report. arXiv preprint arXiv:25…
  3. Shengyuan Chen, Qinggang Zhang, Junnan Dong, Wen Hua, Qing Li, and Xiao Huang. 2024a. Entity alignment with noisy annotations from large language models. In Proceedings of the Neural Information Processing Systems (NeurIPS). find
  4. Yangyi Chen, Karan Sikka, Michael Cogswell, Heng Ji, and Ajay Divakaran. 2024b. DRESS : Instructing large vision-language models to align and interact with humans via natural language feedback. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 14239– 1425… find
  5. Zhe Chen, Weiyun Wang, Hao Tian, Shenglong Ye, Zhangwei Gao, Erfei Cui, Wenwen Tong, Kongzhi Hu, Jiapeng Luo, Zheng Ma, Ji Ma, Jiaqi Wang, Xiaoyi Dong, Hang Yan, Hewei Guo, Conghui He, Botian Shi, Zhenjiang Jin, Chao Xu, and 16 others. 2024c. How far are we to GPT-4V? closing the gap to commercial m…
  6. Zhe Chen, Jiannan Wu, Wenhai Wang, Weijie Su, Guo Chen, Sen Xing, Muyan Zhong, Qinglong Zhang, Xizhou Zhu, Lewei Lu, Bin Li, Ping Luo, Tong Lu, Yu Qiao, and Jifeng Dai. 2024d. InternVL: Scaling up vision foundation models and aligning for generic visual-linguistic tasks. In Proceedings of the IEEE C…
  7. Jianfeng Chi, Ujjwal Karn, Hongyuan Zhan, Eric Smith, Javier Rando, Yiming Zhang, Kate Plawiak, Zacharie Delpierre Coudert, Kartikeya Upasani, and Mahesh Pasupuleti. 2024. Llama guard 3 vision: Safeguarding human-ai image understanding conversations. arXiv preprint arXiv:2411.10414. — [arXiv:2411.10414](https://arxiv.org/abs/2411.10414) find
  8. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the North American Chapter of the Association for Computational Linguistics (NAACL), pages 4171–4186.
  9. Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, and Robin Rombach. 2024. Scaling rectified flow transformers for high-resolution image synthesis. In Forty-firs… find
  10. Team GLM, Aohan Zeng, Bin Xu, Bowen Wang, Chenhui Zhang, Da Yin, Diego Rojas, Guanyu Feng, Hanlin Zhao, Hanyu Lai, Hao Yu, Hongning Wang, Jiadai Sun, Jiajie Zhang, Jiale Cheng, Jiayi Gui, Jie Tang, Jing Zhang, Juanzi Li, and 37 others. 2024. ChatGLM: A family of large language models from GLM-130B t…
  11. Yichen Gong, Delong Ran, Jinyuan Liu, Conglei Wang, Tianshuo Cong, Anyu Wang, Sisi Duan, and Xiaoyun Wang. 2025. FigStep: Jailbreaking large vision-language models via typographic visual prompts. In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), pages 23951–23959. find
  12. Tianle Gu, Zeyang Zhou, Kexin Huang, Liang Dandan, Yixu Wang, Haiquan Zhao, Yuanqi Yao, Yujiu Yang, Yan Teng, Yu Qiao, and Wang Yingchun. 2024. MLLMGuard: A multi-dimensional safety evaluation suite for multimodal large language models. Advances in Neural Information Processing Systems, 37:7256–7295… find
  13. Xuhao Hu, Dongrui Liu, Hao Li, Xuanjing Huang, and Jing Shao. 2024. VLSBench: Unveiling visual leakage in multimodal safety. arXiv preprint arXiv:2411.19939.
  14. Yilei Jiang, Xinyan Gao, Tianshuo Peng, Yingshui Tan, Xiaoyong Zhu, Bo Zheng, and Xiangyu Yue. 2025. HiddenDetect: Detecting jailbreak attacks against large vision-language models via monitoring hidden states. CoRR, abs/2502.14744. — [arXiv:2502.14744](https://arxiv.org/abs/2502.14744) find
  15. Yilei Jiang, Yingshui Tan, and Xiangyu Yue. 2024. RapGuard: Safeguarding multimodal large language models via rationale-aware defensive prompting. CoRR, abs/2412.18826. — [arXiv:2412.18826](https://arxiv.org/abs/2412.18826) find
  16. Black Forest Labs. 2024. Flux. https://github.com/black-forest-labs/flux. ↗
  17. Jian Li, Weiheng Lu, Hao Fei, Meng Luo, Ming Dai, Min Xia, Yizhang Jin, Zhenye Gan, Ding Qi, Tai Ying Yang Wankou Wang Yabiao Fu, Chaoyou, and Wang Chengjie. 2024a. A survey on benchmarks of multimodal large language models. arXiv preprint arXiv:2408.08632. — [arXiv:2408.08632](https://arxiv.org/abs/2408.08632) find
  18. Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. 2023. Blip-2: bootstrapping language-image pre-training with frozen image encoders and large language models. In Proceedings of the 40th International Conference on Machine Learning, pages 19730–19742.
  19. Mukai Li, Lei Li, Yuwei Yin, Masood Ahmed, Zhenguang Liu, and Qi Liu. 2024b. Red teaming visual language models. In Proceedings of the Association for Computational Linguistics (ACL), pages 3326– 3342. find
  20. Xirui Li, Hengguang Zhou, Ruochen Wang, Tianyi Zhou, Minhao Cheng, and Cho-Jui Hsieh. 2024c. MOSSBench: Is your multimodal language model oversensitive to safe queries? CoRR, abs/2406.17806. — [arXiv:2406.17806](https://arxiv.org/abs/2406.17806) find
  21. Yifan Li, Hangyu Guo, Kun Zhou, Wayne Xin Zhao, and Ji-Rong Wen. 2024d. Images are achilles’ heel of alignment: Exploiting visual vulnerabilities for jailbreaking multimodal large language models. In Proceedings of the European Conference on Computer Vision (ECCV), volume 15131, pages 174–189. find
  22. Ji Lin, Hongxu Yin, Wei Ping, Yao Lu, Pavlo Molchanov, Andrew Tao, Huizi Mao, Jan Kautz, Mohammad Shoeybi, and Song Han. 2023. VILA: On pre-training for visual language models.
  23. Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C Lawrence Zitnick. 2014. Microsoft COCO: Common objects in context. In Proceedings of the European Conference on Computer Vision (ECCV), pages 740–755. Springer.
  24. Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2024a. Improved baselines with visual instruction tuning. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 26296–26306.
  25. Xin Liu, Yichen Zhu, Jindong Gu, Yunshi Lan, Chao Yang, and Yu Qiao. 2024b. MM-SafetyBench: A benchmark for safety evaluation of multimodal large language models. In Proceedings of the European Conference on Computer Vision (ECCV), pages 386– 403. find
  26. Zhijian Liu, Ligeng Zhu, Baifeng Shi, Zhuoyang Zhang, Yuming Lou, Shang Yang, Haocheng Xi, Shiyi Cao, Yuxian Gu, Dacheng Li, Xiuyu Li, Yunhao Fang, Yukang Chen, Cheng-Yu Hsieh, De-An Huang, An-Chieh Cheng, Vishwesh Nath, Jinyi Hu, Sifei Liu, and 8 others. 2024c. NVILA: Efficient frontier visual lang… find
  27. Weidi Luo, Siyuan Ma, Xiaogeng Liu, Xiaoyu Guo, and Chaowei Xiao. 2024. Jailbreakv-28k: A benchmark for assessing the robustness of multimodal large language models against jailbreak attacks. CoRR, abs/2404.03027. find
  28. OpenAI. 2023. GPT-4 technical report. arXiv preprint arXiv:2303.08774.
  29. OpenAI. 2024. GPT-4o system card. arXiv preprint arXiv:2410.21276.
  30. Qwen Team. 2025. Qwen3-VL technical report. arXiv preprint arXiv:2511.21631.
  31. Qwen Team. 2026. Qwen3.6: Pushing the limits of language models. https://qwen.ai/blog?id= qwen3.6. ↗
  32. Machel Reid, Nikolay Savinov, Denis Teplyashin, Dmitry Lepikhin, Timothy P. Lillicrap, Jean-Baptiste Alayrac, Radu Soricut, Angeliki Lazaridou, Orhan Firat, Julian Schrittwieser, Ioannis Antonoglou, Rohan Anil, Sebastian Borgeaud, Andrew M. Dai, Katie Millican, Ethan Dyer, Mia Glaese, Thibault Sotti… — Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context
  33. Paul Röttger, Giuseppe Attanasio, Felix Friedrich, Janis Goldzycher, Alicia Parrish, Rishabh Bhardwaj, Chiara Di Bonaventura, Roman Eng, Gaia El Khoury Geagea, Subhra S. Goswami, Jieun Han, Dirk Hovy, Seogyeong Jeong, Paloma Jeretic, Flor Miriam Plaza del Arco, Donya Rooein, Patrick Schramowski, Ana… — [arXiv:2501.10057](https://arxiv.org/abs/2501.10057) find
  34. Christoph Schuhmann, Romain Beaumont, Richard Vencu, Cade Gordon, Ross Wightman, Mehdi Cherti, Theo Coombes, Aarush Katta, Clayton Mullis, Mitchell Wortsman, Patrick Schramowski, Srivatsa Kundurthy, Katherine Crowson, Ludwig Schmidt, Robert Kaczmarczyk, and Jenia Jitsev. 2022. LAION-5N: An open larg… find
  35. Chen Shengyuan, Yunfeng Cai, Huang Fang, Xiao Huang, and Mingming Sun. 2023. Differentiable neuro-symbolic reasoning on large-scale knowledge graphs. In Proceedings of the Neural Information Processing Systems (NeurIPS), volume 36. find
  36. Charles Spearman. 1961. The proof and measurement of association between two things. find
  37. Yingshui Tan, Yilei Jiang, Yanshi Li, Jiaheng Liu, Xingyuan Bu, Wenbo Su, Xiangyu Yue, Xiaoyong Zhu, and Bo Zheng. 2025. Equilibrate RLHF: Towards balancing helpfulness-safety trade-off in large language models. arXiv preprint arXiv:2502.11555. — [arXiv:2502.11555](https://arxiv.org/abs/2502.11555) find
  38. Yingshui Tan, Boren Zheng, Baihui Zheng, Kerui Cao, Huiyun Jing, Jincheng Wei, Jiaheng Liu, Yancheng He, Wenbo Su, Xiaoyong Zhu, Bo Zheng, and Kaifu Zhang. 2024. Chinese SafetyQA: A safety short-form factuality benchmark for large language models. arXiv preprint arXiv:2412.15265.
  39. Haoqin Tu, Chenhang Cui, Zijun Wang, Yiyang Zhou, Bingchen Zhao, Junlin Han, Wangchunshu Zhou, Huaxiu Yao, and Cihang Xie. 2024. How many are in this image a safety evaluation benchmark for vision LLMs. In Proceedings of the European Conference on Computer Vision (ECCV), pages 37–55. find
  40. Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Yang Fan, Kai Dang, Mengfei Du, Xuancheng Ren, Rui Men, Dayiheng Liu, Chang Zhou, Jingren Zhou, and Junyang Lin. 2024a. Qwen2-VL: Enhancing vision-language model’s perception of the…
  41. Weihan Wang, Qingsong Lv, Wenmeng Yu, Wenyi Hong, Ji Qi, Yan Wang, Junhui Ji, Zhuoyi Yang, Lei Zhao, Xixuan Song, Jiazheng Xu, Bin Xu, Juanzi Li, Yuxiao Dong, Ming Ding, and Jie Tang. 2023. CogVLM: Visual expert for pretrained language models.
  42. Wenxuan Wang, Xiaoyuan Liu, Kuiyi Gao, Jen-tse Huang, Youliang Yuan, Pinjia He, Shuai Wang, and Zhaopeng Tu. 2025. Can’t see the forest for the trees: Benchmarking multimodal safety awareness for multimodal llms. In Proceedings of the Association for Computational Linguistics (ACL), pages 16993–1700… find
  43. Yu Wang, Xiaofei Zhou, Yichen Wang, Geyuan Zhang, and Tianxing He. 2024b. Jailbreak large vision-language models through multi-modal linkage. arXiv preprint arXiv:2412.00473.
  44. Qi Xue, Minrui Jiang, Runjia Zhang, Xiurui Xie, Pei Ke, and Guisong Liu. 2025. Falcon: A cross-modal evaluation dataset for comprehensive safety perception. arXiv preprint arXiv:2509.23783. — [arXiv:2509.23783](https://arxiv.org/abs/2509.23783) find
  45. Yuan Yao, Tianyu Yu, Ao Zhang, Chongyi Wang, Junbo Cui, Hongji Zhu, Tianchi Cai, Haoyu Li, Weilin Zhao, Zhihui He, Qianyu Chen, Huarong Zhou, Zhensheng Zou, Haoye Zhang, Shengding Hu, Zhi Zheng, Jie Zhou, Jie Cai, Xu Han, and 4 others. 2024. MiniCPM-V: A GPT-4V level MLLM on your phone. arXiv prepri…
  46. Mang Ye, Xuankun Rong, Wenke Huang, Bo Du, Nenghai Yu, and Dacheng Tao. 2025. A survey of safety on large vision-language models: Attacks, defenses and evaluations. arXiv preprint arXiv:2502.14881. — [arXiv:2502.14881](https://arxiv.org/abs/2502.14881) find
  47. Zonghao Ying, Aishan Liu, Siyuan Liang, Lei Huang, Jinyang Guo, Wenbo Zhou, Xianglong Liu, and Dacheng Tao. 2024. SafeBench: A safety evaluation framework for multimodal large language models. arXiv preprint arXiv:2410.18927. — [arXiv:2410.18927](https://arxiv.org/abs/2410.18927) find
  48. Jiahao Yu, Xingwei Lin, Zheng Yu, and Xinyu Xing. 2024. LLM-Fuzzer: Scaling assessment of large language model jailbreaks. In 33rd USENIX Security Symposium (USENIX Security 24), pages 4657–4674. find
  49. Tongxin Yuan, Zhiwei He, Lingzhong Dong, Yiming Wang, Ruijie Zhao, Tian Xia, Lizhen Xu, Binglin Zhou, Fangqi Li, Zhuosheng Zhang, Rui Wang, and Gongshen Liu. 2024. R-judge: Benchmarking safety risk awareness for LLM agents. arXiv preprint arXiv:2401.10019.
  50. Hao Zhang, Wenqi Shao, Hong Liu, Yongqiang Ma, Ping Luo, Yu Qiao, and Kaipeng Zhang. 2024a. AVIBench: Towards evaluating the robustness of large vision-language model on adversarial visual-instructions. CoRR, abs/2403.09346. — [arXiv:2403.09346](https://arxiv.org/abs/2403.09346) find
  51. Qinggang Zhang, Shengyuan Chen, Yuanchen Bei, Zheng Yuan, Huachi Zhou, Zijin Hong, Junnan Dong, Hao Chen, Yi Chang, and Xiao Huang. 2025a. A survey of graph retrieval-augmented generation for customized large language models. arXiv preprint arXiv:2501.13958.
  52. Yichi Zhang, Yao Huang, Yitong Sun, Chang Liu, Zhe Zhao, Zhengwei Fang, Yifan Wang, Huanran Chen, Xiao Yang, Xingxing Wei, Hang Su, Yinpeng Dong, and Jun Zhu. 2024b. Benchmarking trustworthiness of multimodal large language models: A comprehensive study. CoRR, abs/2406.07057. — [arXiv:2406.07057](https://arxiv.org/abs/2406.07057) find
  53. Yongting Zhang, Lu Chen, Guodong Zheng, Yifeng Gao, Rui Zheng, Jinlan Fu, Zhenfei Yin, Senjie Jin, Yu Qiao, Xuanjing Huang, Feng Zhao, Tao Gui, and Jing Shao. 2024c. SPA-VL:A comprehensive safety preference alignment dataset for vision language model. CoRR, abs/2406.12030. — [arXiv:2406.12030](https://arxiv.org/abs/2406.12030) find
  54. Zhexin Zhang, Leqi Lei, Lindong Wu, Rui Sun, Yongkang Huang, Chong Long, Xiao Liu, Xuanyu Lei, Jie Tang, and Minlie Huang. 2023. Safety-Bench: Evaluating the safety of large language models. arXiv preprint arXiv:2309.07045.
  55. Ziyi Zhang, Zhen Sun, Zongmin Zhang, Jihui Guo, and Xinlei He. 2025b. Fc-attack: Jailbreaking large vision-language models via auto-generated flowcharts. arXiv preprint arXiv:2502.21059. — [arXiv:2502.21059](https://arxiv.org/abs/2502.21059) find
  56. Shiji Zhao, Ranjie Duan, Fengxiang Wang, Chi Chen, Caixin Kang, Jialing Tao, YueFeng Chen, Hui Xue, and Xingxing Wei. 2025. Jailbreaking multimodal large language models via shuffle inconsistency. arXiv preprint arXiv:2501.04931. — [arXiv:2501.04931](https://arxiv.org/abs/2501.04931) find
  57. Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, Yifan Du, Chen Yang, Yushuo Chen, Zhipeng Chen, Jinhao Jiang, Ruiyang Ren, Yifan Li, Xinyu Tang, Zikang Liu, and 3 others. 2023. A survey of large language models. arXiv…
  58. Kaiwen Zhou, Chengzhi Liu, Xuandong Zhao, Anderson Compalas, Dawn Song, and Xin Eric Wang. 2024. Multimodal situational safety. CoRR, abs/2410.06172. — [arXiv:2410.06172](https://arxiv.org/abs/2410.06172) find
  59. Yongshuo Zong, Ondrej Bohdal, Tingyang Yu, Yongxin Yang, and Timothy Hospedales. 2024. Safety fine-tuning at (almost) no cost: A baseline for vision large language models. In Proceedings of the International Conference on Machine Learning (ICML), pages 62867–62891. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zheng-2026-usb-comprehensive-unified-safety-evaluation/