I am a joint PhD student at Fudan University and Shanghai Innovation Institute (SII).
My research focuses on Trustworthy AI, AI Safety, and Agent Safety, with particular interests in building safer and more controllable LLM and agentic systems.
我目前主要研究 大语言模型与 Agent 系统安全，重点关注 Agent Guard、LLM Guard、运行时安全防护 以及 面向治理机制的安全控制方法。
| Area | Description |
|---|---|
| Agent Guard | Safety mechanisms for constraining risky Agent behaviors |
| LLM Guard | Guard strategies for LLMs, protecting both inputs and outputs against unsafe or harmful content |
| Runtime Safety | Monitoring, intervention, and protection during execution |
| Risk Testing | Stress-testing safety boundaries of AI systems |
| Governance-inspired Defense | Approval, auditing, and control mechanisms for trustworthy deployment |
- Safety for LLMs and Agents
- Agent Guard and LLM Guard
- Risk Testing and Path Protection
Research Mode: ACTIVE
Primary Direction: Agent Safety / LLM Safety
Keywords: Agent Guard, LLM Guard, Runtime Protection
Current Goal: Build safer, more trustworthy, and more controllable AI systems