AI safety researcher

Haojin Zhu

Shanghai Jiao Tong University

Google Scholar dblp ORCID

Last 12 months6 papers

Attacks 2

AI safety papers from the last year

On this site: 17 papers

July 2026

Security of World-Model-Based Embodied AI: A Lifecycle of Threats, Defenses, and Evaluation
Jul 30arXiv ↗

June 2026

DIPBox: A Multi-scale Testing Framework for Tracking Dataset Regeneration
Jun 19arXiv ↗

March 2026

Trojan's Whisper: Stealthy Manipulation of OpenClaw through Injected Bootstrapped Guidance
Mar 20AttacksarXiv ↗
EmbTracker: Traceable Black-box Watermarking for Federated Language Models
Mar 12arXiv ↗
SlowBA: An efficiency backdoor attack towards VLM-based GUI agents
Mar 9AttacksarXiv ↗

December 2025

Patronus: Identifying and Mitigating Transferable Backdoors in Pre-trained Language Models
Dec 7arXiv ↗

Authorship follows dblp and the author's own Google Scholar profile; only AI safety papers on this site are listed.