Huazhong University of Science and Technology and Alibaba have launched an AI security research center concentrating on key areas such as large model security and intelligent system protection. The partnership kicked off nine initial projects targeting AI security challenges including model attack defense, endpoint safety, and data privacy.
Focus Areas of the AI Security Research Center
The newly formed center emphasizes four main domains: large model security offense and defense, intelligent agent and endpoint protection, smart system and software security, and data collaboration with privacy safeguards. The first batch of nine projects covers these critical topics to fortify AI capabilities against emerging threats.
Highlighted Research Themes
- Image Adversarial Samples and Endpoint Threat Assessment: With AI agents operating users’ smartphones and performing complex tasks, there is a risk of screen forgeries and malicious manipulation. This project aims to develop techniques to identify and counter such image-based security threats.
- Privacy Protection Relationship Graph: Users often switch between apps like shopping, maps, and food delivery, expecting seamless data flow without risking personal information leaks. This initiative explores privacy-preserving methods such as secure computing to enable data usability without exposure, balancing convenience with privacy.









