Lijun Li

Lijun Li

Researcher at Shanghai AI Lab

Hi there! I am now an Researcher at Shanghai AI Lab. Before joining Shanghai AI Lab, I was a Researcher at Machine Intelligence Technology Lab, DAMO Academy, Alibaba Group, under the guidance of Rong Jin and Ping Tan. I have been engaged in AI research and development since 2019 after PHD graduation. I am fortunate to have gained industry experience at Alibaba DAMO Academy, Tencent AI Lab, and Baidu IDL.

NOTE: We are hiring full-time Researcher and Engineer, motivated interns and joint-training PhD students. If you are enthusiastic about Safe AI and wish to join us, contact me at lilijun@pjlab.org.cn directly!

If interested in collaboration or discussion, please email me.

πŸ”­ Research Interest

My research interests lie at the nexus of AI safety and visual understanding. I aim to develop artificial general intelligence with built‑in, balanced safety capabilities. To this end, I concentrate on thoroughly characterizing AI safety vulnerabilities，and embedding robust protective measures from the very outset of model design and training.

πŸ”₯ News

  • 2026.07: Our paper EGD is awarded as ACL 2026 Outstanding paper.
  • 2026.06: Our papers CIA has been accepted by ECCV 2026.
  • 2026.05： Our papers OpenSafeRL and MemGuard have been accepted by ICML 2026.
  • 2026.04: πŸŽ‰ Our papers EGD, SEARL, HarmRLVR and ToolSafe have been accepted by ACL 2026.
  • 2026.02οΌšπŸŽ‰ Our paper TreeTeaming has been accepted by CVPR 2026.
  • 2026.01: πŸŽ‰ Our papers MIS and GhostEI-Bench have been accepted by ICLR 2026.
  • 2026.01: πŸŽ‰ We release a comprehensive survey paper about Efficient Agent.
  • 2025.11οΌšπŸŽ‰ Our paper Response Attack has been accepted by AAAI 2026.
  • 2025.09: πŸŽ‰ Our paper PURE has been accepted by Neurips 2025.
  • 2025.08: πŸŽ‰ Our papers LARF, Visco-attack and RMS have been accepted by EMNLP 2025.
  • 2025.07: πŸŽ‰ We release a safety-enhanced framework SafeWork-R1 for VLM.
  • 2025.05: πŸŽ‰ Our paper WorldSimBench has been accepted by ICML 2025.
  • 2025.02: πŸŽ‰ Our paper T2ISafety has been accepted by CVPR 2025.
  • 2024.05: πŸŽ‰ Our papers SALAD-Bench and PsySafe have been accepted by ACL 2024. PsySafe is awarded as Outstanding paper.
  • 2024.03: πŸŽ‰ We release an easy-to-use Python framework to generate adversarial jailbreak prompts at website.
  • 2023: πŸŽ‰ Our paper RenderIH has been accepted by ICCV 2023.

  • 2022.05: πŸŽ‰πŸŽ‰ We rank 1st at the Action Detection track of EPIC@CVPR2022 Workshop

  • 2022.04οΌšπŸŽ‰ We rank 2nd at the FreiHAND Competition

πŸ“ Publications

Topics: AI Safety/ Agent / Understanding

(*: indicates equal contribution; ‑: indicates corresponding)

ACL 2026
sym

Evolutionary Guided Decoding: Iterative Value Refinement for LLMs

Zhenhua Liu*, Lijun Li*, Ruizhe Chen, Yuxian Jiang, Tong Zhu, Zhaochen Su, Wenliang Chen, Jing Shao‑

ACL 2026

Paper

ACL 2026
sym

HarmRLVR: Weaponizing Verifiable Rewards for Harmful LLM Alignment

Yuexiao Liu*, Lijun Li*‑, Xingjun Wang, Jing Shao‑

ACL 2026

Paper Code

CVPR 2026
sym

TreeTeaming: Autonomous Red-Teaming of Vision-Language Models via Hierarchical Strategy Exploration

Chunxiao Li, Lijun Li‑, Jing Shao‑

CVPR 2026

Paper

Preprint
sym

Toward Efficient Agents: A Survey of Memory, Tool learning, and Planning

Xiaofang Yang, Lijun Li‑, Heng Zhou, Tong Zhu, Xiaoye Qu, Yuchen Fan, Qianshan Wei, Rui Ye, Li Kang, Yiran Qin, Zhiqiang Kou, Daizong Liu, Qi Li, Ning Ding, Siheng Chen, Jing Shao‑

Preprint, 2026

Paper Project Code

ACL 2026 Findings
sym

ToolSafe: Enhancing Tool Invocation Safety of LLM-based agents via Proactive Step-level Guardrail and Feedback

Yutao Mou, Zhangchi Xue, Lijun Li‑, Peiyang Liu, Shikun Zhang, Wei Ye‑, Jing Shao‑

ACL 2026

Paper Code Guard

Preprint
sym

ProGuard: Towards Proactive Multimodal Safeguard

Shaohan Yu*, Lijun Li*‑, Chenyang Si, Lu Sheng, Jing Shao‑

Preprint, 2025

Paper Code Guard-3B Guard-7B Dataset

ECCV 2026
sym

Contextual Image Attack: How Visual Context Exposes Multimodal Safety Vulnerabilities

Yuan Xiong*, Ziqi Miao*, Lijun Li*‑, Chen Qian, Jie Li, Jing Shao‑

ECCV 2026

Paper Code

Preprint
sym

STaR-Attack: A Spatio-Temporal and Narrative Reasoning Attack Framework for Unified Multimodal Understanding and Generation Models

Shaoxiong Guo*, Tianyi Du*, Lijun Li*‑, Yuyao Wu, Jie Li, Jing Shao‑

Preprint, 2025

Paper

Preprint
sym

Collaborative Shadows: Distributed Backdoor Attacks in LLM-Based Multi-Agent Systems

Pengyu Zhu*, Lijun Li*‑, Yaxing Lyu*, Li Sun, Sen Su‑, Jing Shao‑

Preprint, 2025

Paper Code

ICLR 2026
sym

Rethinking bottlenecks in safety fine-tuning of vision language models

Yi Ding*, Lijun Li*, Bing Cao‑, Jing Shao‑

ICLR 2026

Paper Project Data

Tech report
sym

SafeWork-R1: Coevolving Safety and Intelligence under the AI-45β—¦ Law

Worked as Co-Leads

Tech report, 2025

Paper

EMNLP 2025 Findings
sym

Self-adaptive Dataset Construction for Real-World Multimodal Safety Scenarios

Jingen Qu*, Lijun Li*‑, Bo Zhang, Yichen Yan, Jing Shao‑

EMNLP 2025 Findings

Paper

AAAI 2026
sym

Response Attack: Exploiting Contextual Priming to Jailbreak Large Language Models

Ziqi Miao*, Lijun Li*‑, Yuan Xiong, Zhenhua Liu, Pengyu Zhu, Jing Shao‑

AAAI, 2026

Paper Data

EMNLP 2025
sym

Layer-Aware Representation Filtering: Purifying Finetuning Data to Preserve LLM Safety Alignment

Hao Li*, Lijun Li*‑, Zhenghao Lu, Xianyi Wei, Rui Li, Jing Shao‑, Lei Sha‑

EMNLP 2025

Paper Code

EMNLP 2025
sym

Visual Contextual Attack: Jailbreaking MLLMs with Image-Driven Context Injection

Ziqi Miao*, Yi Ding*, Lijun Li‑, Jing Shao‑

EMNLP 2025

Paper Data

ICML 2025
sym

Worldsimbench: Towards video generation models as world simulators

Yiran Qin*, Zhelun Shi*, Jiwen Yu, Xijun Wang, Enshen Zhou, Lijun Li, Zhenfei Yin, Xihui Liu, Lu Sheng, Jing Shao‑, Lei Bai‑, Wanli Ouyang, Ruimao Zhang‑

ICML 2025

Paper Project

CVPR 2025
sym

T2isafety: Benchmark for assessing fairness, toxicity, and privacy in image generation

Lijun Li*, Zhelun Shi*, Xuhao Hu, Bowen Dong, Yiran Qin, Xihui Liu, Lu Sheng, Jing Shao‑

CVPR 2025

Paper Data ImageGuard

ACL 2024
sym

Psysafe: A comprehensive framework for psychological-based attack, defense, and evaluation of multi-agent system safety

Zaibin Zhang, Yongting Zhang, Lijun Li, Hongzhi Gao, Lijun Wang, Huchuan Lu, Feng Zhao, Yu Qiao, Jing Shao

πŸŽ–οΈACL 2024 Outstanding Paper

Paper Code

ACL 2024 Findings
sym

SALAD-Bench: A Hierarchical and Comprehensive Safety Benchmark for Large Language Models

Lijun Li*, Bowen Dong*, Ruohui Wang*, Xuhao Hu*, Wangmeng Zuo, Dahua Lin, Yu Qiao, Jing Shao‑

ACL 2024 Findings

Paper Project Code Data MD-Judge Leaderboard

Preprint
sym

EasyJailbreak: A Unified Framework for Jailbreaking Large Language Models

Weikang Zhou, Xiao Wang, Limao Xiong, Han Xia, Yingshuang Gu, Mingxu Chai, Fukang Zhu, Caishuang Huang, Shihan Dou, Zhiheng Xi, Rui Zheng, Songyang Gao, Yicheng Zou, Hang Yan, Yifan Le, Ruohui Wang, Lijun Li, Jing Shao, Tao Gui, Qi Zhang, Xuanjing Huang

Preprint, 2024

Paper Project Code

Preprint
sym

Assessment of multimodal large language models in alignment with human values

Zhelun Shi, Zhipin Wang, Hongxing Fan, Zaibin Zhang, Lijun Li, Yongting Zhang, Zhenfei Yin, Lu Sheng, Yu Qiao, Jing Shao

Preprint, 2024

Paper Project Code

ICCV 2023
sym

Renderih: A large-scale synthetic dataset for 3d interacting hand pose estimation

Lijun Li‑, Linrui Tian, Xindi Zhang, Qi Wang, Bang Zhang, Liefeng Bo, Mengyuan Liu, Chen Chen

ICCV 2023

Paper Code Imgs Annotations

Preprint
sym

One-stage Action Detection Transformer

Lijun Li, Li’an Zhuo, Bang Zhang

πŸŽ–οΈWinning solution for Action Detection Track of EPIC@CVPR2022

Paper

ICML 2019
sym

Nattack: Learning the distributions of adversarial examples for an improved black-box attack on deep neural networks

Yandong Li*, Lijun Li*, Liqiang Wang, Tong Zhang, Boqing Gong

ICML 2019

Paper Code

πŸ–ŠοΈ Research Experiences

  • 2023 - Present, Researcher, Shanghai AI Lab, Beijing, China.
  • 2021 - 2023, Researcher, DAMO Academy, Alibaba, Beijing, China.

πŸŽ– Honors and Awards

πŸ’¬ Invited Talks

πŸ‘¨β€πŸ”§ Academic Services