DeepSight: An All-in-One LM Safety Toolkit
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866915794911232000 |
|---|---|
| author | Zhang, Bo Guo, Jiaxuan Li, Lijun Liu, Dongrui Chen, Sujin Chen, Guanxu Zheng, Zhijie Lin, Qihao Yan, Lewen Qian, Chen Zhou, Yijin Wu, Yuyao Guo, Shaoxiong Du, Tianyi Yang, Jingyi Hu, Xuhao Miao, Ziqi Lu, Xiaoya Shao, Jing Hu, Xia |
| author_facet | Zhang, Bo Guo, Jiaxuan Li, Lijun Liu, Dongrui Chen, Sujin Chen, Guanxu Zheng, Zhijie Lin, Qihao Yan, Lewen Qian, Chen Zhou, Yijin Wu, Yuyao Guo, Shaoxiong Du, Tianyi Yang, Jingyi Hu, Xuhao Miao, Ziqi Lu, Xiaoya Shao, Jing Hu, Xia |
| contents | As the development of Large Models (LMs) progresses rapidly, their safety is also a priority. In current Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) safety workflow, evaluation, diagnosis, and alignment are often handled by separate tools. Specifically, safety evaluation can only locate external behavioral risks but cannot figure out internal root causes. Meanwhile, safety diagnosis often drifts from concrete risk scenarios and remains at the explainable level. In this way, safety alignment lack dedicated explanations of changes in internal mechanisms, potentially degrading general capabilities. To systematically address these issues, we propose an open-source project, namely DeepSight, to practice a new safety evaluation-diagnosis integrated paradigm. DeepSight is low-cost, reproducible, efficient, and highly scalable large-scale model safety evaluation project consisting of a evaluation toolkit DeepSafe and a diagnosis toolkit DeepScan. By unifying task and data protocols, we build a connection between the two stages and transform safety evaluation from black-box to white-box insight. Besides, DeepSight is the first open source toolkit that support the frontier AI risk evaluation and joint safety evaluation and diagnosis. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_12092 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | DeepSight: An All-in-One LM Safety Toolkit Zhang, Bo Guo, Jiaxuan Li, Lijun Liu, Dongrui Chen, Sujin Chen, Guanxu Zheng, Zhijie Lin, Qihao Yan, Lewen Qian, Chen Zhou, Yijin Wu, Yuyao Guo, Shaoxiong Du, Tianyi Yang, Jingyi Hu, Xuhao Miao, Ziqi Lu, Xiaoya Shao, Jing Hu, Xia Computation and Language Artificial Intelligence Cryptography and Security Computer Vision and Pattern Recognition As the development of Large Models (LMs) progresses rapidly, their safety is also a priority. In current Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) safety workflow, evaluation, diagnosis, and alignment are often handled by separate tools. Specifically, safety evaluation can only locate external behavioral risks but cannot figure out internal root causes. Meanwhile, safety diagnosis often drifts from concrete risk scenarios and remains at the explainable level. In this way, safety alignment lack dedicated explanations of changes in internal mechanisms, potentially degrading general capabilities. To systematically address these issues, we propose an open-source project, namely DeepSight, to practice a new safety evaluation-diagnosis integrated paradigm. DeepSight is low-cost, reproducible, efficient, and highly scalable large-scale model safety evaluation project consisting of a evaluation toolkit DeepSafe and a diagnosis toolkit DeepScan. By unifying task and data protocols, we build a connection between the two stages and transform safety evaluation from black-box to white-box insight. Besides, DeepSight is the first open source toolkit that support the frontier AI risk evaluation and joint safety evaluation and diagnosis. |
| title | DeepSight: An All-in-One LM Safety Toolkit |
| topic | Computation and Language Artificial Intelligence Cryptography and Security Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2602.12092 |