Guanyu Hou

Hello, I’m Guanyu Hou. I am an MSc in Artificial Intelligence student at The University of Manchester. My research focuses on AI safety and security, with an emphasis on trustworthy large language models and multimodal systems.

I study the robustness of audio language models, safety evaluation of text-to-video models, backdoor vulnerabilities, and privacy in retrieval-augmented generation. My work has appeared at venues including CVPR, EMNLP, AAAI, ACM Multimedia Asia, ICASSP, and ICME.

I am interested in PhD opportunities in AI safety and security.

Publications · CV · Google Scholar · Email

Internship Experience

Shanghai Artificial Intelligence Laboratory (PJLab) — Research Intern, Sep 2026–Present. Working with Dr. Lijun Li on AI4Science safety.
ByteDance — Algorithm Intern (Intelligent Security), Jul–Sep 2026. Worked on multilingual safety classifiers, long-reasoning optimization, and jailbreak defense.

See my CV for details.

Academic Service

AAAI 2027 — Program Committee Member
NeurIPS 2026 — Reviewer
AAAI 2026 — Reviewer

selected publications

arXiv
ChronoLock: Protecting Videos from Unauthorized Text-to-Video Personalization

Jiaming He, Jiashu Zhang, Guanyu Hou, Shuhan Ye, Hanwei Zhu, Yi Yu, and Xudong Jiang

arXiv preprint arXiv:2606.21146, Jun 2026

Abs DOI arXiv Bib HTML PDF

ChronoLock proactively protects shared videos from unauthorized text-to-video personalization. It applies bounded perturbations to temporal denoising trajectories to disrupt motion learning within video chunks and continuity across chunk boundaries. Transformation-sampled optimization improves resilience to common preprocessing, and evaluation on UCF Sports and HMDB51 measures the reduction in motion imitation.
@article{he2026chronolock, title = {ChronoLock: Protecting Videos from Unauthorized Text-to-Video Personalization}, author = {He, Jiaming and Zhang, Jiashu and Hou, Guanyu and Ye, Shuhan and Zhu, Hanwei and Yu, Yi and Jiang, Xudong}, year = {2026}, month = jun, journal = {arXiv preprint arXiv:2606.21146}, doi = {10.48550/arXiv.2606.21146}, }

CVPR

Oral TEAR: Temporal-aware Automated Red-teaming for Text-to-Video Models

Jiaming He, Guanyu Hou, Hongwei Li, Zhicong Huang, Kangjie Chen, Yi Yu, Wenbo Jiang, Guowen Xu, and Tianwei Zhang

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Jun 2026

arXiv Bib HTML PDF

@inproceedings{he2026tear,
  title = {TEAR: Temporal-aware Automated Red-teaming for Text-to-Video Models},
  author = {He, Jiaming and Hou, Guanyu and Li, Hongwei and Huang, Zhicong and Chen, Kangjie and Yu, Yi and Jiang, Wenbo and Xu, Guowen and Zhang, Tianwei},
  year = {2026},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = jun,
  pages = {41--50}
}

MMAsia

When Hallucinated Concepts Cross Modals: Unveiling Backdoor Vulnerability in Multi-modal In-context Learning

Guanyu Hou, Jiaming He, Yitong Qiao, Jiachen Li, Qiyang Song, Ji Guo, Zihan Wang, and Wenbo Jiang

ACM International Conference on Multimedia in Asia (MMAsia), Dec 2025

DOI Bib HTML

@inproceedings{hou2025hallucinated,
  title = {When Hallucinated Concepts Cross Modals: Unveiling Backdoor Vulnerability in Multi-modal In-context Learning},
  author = {Hou, Guanyu and He, Jiaming and Qiao, Yitong and Li, Jiachen and Song, Qiyang and Guo, Ji and Wang, Zihan and Jiang, Wenbo},
  year = {2025},
  booktitle = {Proceedings of the 7th ACM International Conference on Multimedia in Asia},
  doi = {10.1145/3743093.3770938},
  month = dec,
  pages = {1--7}
}

EMNLP

Evaluating Robustness of Large Audio Language Models to Audio Injection: An Empirical Study

Guanyu Hou^*, Jiaming He^*, Yinhang Zhou, Ji Guo, Yitong Qiao, Rui Zhang, and Wenbo Jiang

* Equal contribution.

Conference on Empirical Methods in Natural Language Processing (EMNLP), Nov 2025

DOI Bib HTML PDF

@inproceedings{hou2025audio,
  title = {Evaluating Robustness of Large Audio Language Models to Audio Injection: An Empirical Study},
  author = {Hou, Guanyu and He, Jiaming and Zhou, Yinhang and Guo, Ji and Qiao, Yitong and Zhang, Rui and Jiang, Wenbo},
  year = {2025},
  booktitle = {Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing},
  doi = {10.18653/v1/2025.emnlp-main.1303},
  pages = {25660--25676},
  month = nov,
}