Jindong Gu

I am a Senior Research Scientist at Google, specializing in advancing the reliability and safety of AI technologies. I am also affiliated with the University of Oxford as a Senior Research Fellow. Previously, I focused on ensuring the safety of foundation models within the Gemini Safety team of DeepMind. Currently, my work centers on enhancing the reliability and robustness of AI agents in CAIR team.

I received my Ph.D. in 2022 from the Tresp Lab at the University of Munich, Germany, advised by Prof. Volker Tresp. Following this, I worked as a Postdoctoral Researcher in the Torr Vision Group, led by Prof. Philip Torr, at the University of Oxford, UK, in 2023.

My research goal is to build Responsible AI. Specifically, I am interested in the interpretability, robustness, privacy, and safety of

Visual Perception,
Foundation Model-based Understanding and Reasoning,
Robotic Policy and Planning,
and their Fusion towards General Intelligence Systems.

[Hiring!] If you like to work with me as a Research Intern or FT Scientist in Google, please drop me an email, or contact me on LinkedIn or X. I speak English, Chinese, and German.

News

12 / 2024: Two workshop proposals (Test-time Scaling for CV and Eval-FoMo) accepted in CVPR 2025.
11 / 2024: Invited Talk at PFATCV Workshop and MediaTrust Workshop in BMVC 2024.
10 / 2024: I co-organize Eval-FoMo Workshop and give a talk in GigaVision Workshop in ECCV 2024.
07 / 2024: Six papers on AI Safety & Privacy are accepted to ECCV 2024.
07 / 2024: Papers on AI Safety & Interpretability are accepted to ICML, COLM, and NeurIPS 2024.
06 / 2024: I will serve as Area Chair in NeurIPS 2024 Datasets and Benchmarks Track.
04 / 2024: Our survey paper on Adversarial Transferability is accepted to TMLR.
02 / 2024: Three papers on AI safety are accepted to CVPR 2024.
01 / 2024: Three papers on AI safety (two Spotlight) are accepted to ICLR 2024.
01 / 2024: Two Journal papers on AI safety is accepted to TIFS.

Previous News

12 / 2023: Three papers on AI safety and robustness are accepted to AAAI 2024.
11 / 2023: I started a part-time Faculty Scientist position at Google DeepMind.
10 / 2023: I was promoted to Senior Research Fellow at University of Oxford.
09 / 2023: One paper is accepted to BMVC 2023 and One Benchmark paper to NeurIPS 2023.
07 / 2023: Three papers on VLM Understanding and Robustness are accepted to ICCV 2023.
05 / 2023: Selected as CVPR 2023 Outstanding Reviewer.
05 / 2023: One paper has been accepted to ACL Findings 2023.
02 / 2023: One paper has been accepted to CVPR 2023.
11 / 2022: I join Torr Vision Group as Postdoctoral Researcher at University of Oxford.
08 / 2022: I join Google Responsible ML Team as a Research Intern.
07 / 2022: Four papers on Robustness of Vision Systems are accepted to ECCV 2022.
04 / 2021: I join Microsoft Research Asia as a Research Intern.
03 / 2021: One paper has been accepted as oral to CVPR 2021.
01 / 2021: One paper has been accepted to ICLR 2021.

Experience

11 / 2023 - present: Faculty Scientist at Google, New York, USA
10 / 2023 - present: Senior Researcher at University of Oxford, Oxford, UK
11 / 2022 - 09 / 2023: Postdoctoral Researcher at University of Oxford, Oxford, UK
08 / 2022 - 11 / 2022: Research Intern at Google, New York, USA
04 / 2021 - 02 / 2022: Research Intern at Microsoft Research Asia, Beijing, China
08 / 2020 - 03 / 2021: Research Intern at Tencent AI Lab, Shenzhen, China
09 / 2017 - 08 / 2020: Doctoral Researcher at Siemens Technology, Munich, Germany

Selected Publications

Full Publication List

- Responsible Generative AI: What to Generate and What Not [PDF]
  
  Jindong Gu
  
  Under Review, 2024
- Self-Discovering Interpretable Diffusion Latent Directions for Responsible Text-to-Image Generation [PDF][CODE]
  
  Hang Li, Chengzhi Shen, Philip Torr, Volker Tresp, Jindong Gu
  
  IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2024
- An Image Is Worth 1000 Lies: Transferability of Adversarial Images across Prompts on Vision-Language Models [PDF][CODE]
  
  Haochen Luo*, Jindong Gu*, Fengyuan Liu, Philip Torr
  
  International Conference on Learning Representations (ICLR) , 2024
- Are Vision Transformers robust to Patch-wise Perturbation? [PDF][CODE]
  
  Jindong Gu, Volker Tresp, Yao Qin
  
  European Conference on Computer Vision (ECCV) , 2022
- SegPGD: An Effective and Efficient Adversarial Attack for Evaluating and Boosting Segmentation Robustness [PDF]
  
  Jindong Gu, Hengshuang Zhao, Volker Tresp, Phillip Torr
  
  European Conference on Computer Vision (ECCV) , 2022