HUST and Alibaba Jointly Establish AI Frontier Security Research Center, Covering Large Model Security Offense and Defense, Among Others
IT Home, November 21 — Huazhong University of Science and Technology (HUST) and Alibaba Group signed a cooperation agreement on November 18 to jointly build the "Intelligent Offense-Defense and Frontier Security Joint Research Center" focused on AI security, with 9 research projects launched in the first batch.

IT Home has learned that the 9 projects launched in the first batch by the joint research center cover four major directions: large model security offense and defense, agent and endpoint security, intelligent systems and software security, and data collaboration and privacy protection. The following are the cases officially disclosed:
When AI agents begin to operate phones and complete complex tasks on behalf of users, screen images may be forged, and the operation process may also be maliciously induced. In response, the joint research center has established the "Image Adversarial Samples and Endpoint Threat Assessment" project to explore identification and protection technologies for related security risks. When users switch between applications such as shopping, maps, and food delivery, they want to enjoy the convenience brought by data interoperability while worrying about personal information leaking during collaboration. To address this issue, the "Privacy-Preserving Relationship Graph" project will explore technologies such as confidential computing to achieve "data usable but not visible, collaborative but not passed through," so that convenience and privacy protection no longer become a single-choice question.
When AI agents begin to operate phones and complete complex tasks on behalf of users, screen images may be forged, and the operation process may also be maliciously induced. In response, the joint research center has established the "Image Adversarial Samples and Endpoint Threat Assessment" project to explore identification and protection technologies for related security risks.
When users switch between applications such as shopping, maps, and food delivery, they want to enjoy the convenience brought by data interoperability while worrying about personal information leaking during collaboration. To address this issue, the "Privacy-Preserving Relationship Graph" project will explore technologies such as confidential computing to achieve "data usable but not visible, collaborative but not passed through," so that convenience and privacy protection no longer become a single-choice question.
Related Articles
- Alibaba's net profit attributable to parent for the first half of fiscal year 2027 was 75.072 billion yuan, up 17% year on year
- Report says Alibaba's B2B business is being integrated domestically and internationally, with Zhang Kuo overseeing both 1688 and Alibaba.com
- Alibaba's net profit attributable to parent for the first quarter of fiscal year 2027 was 10.537 billion yuan, down 76% year on year
- Alibaba Qwen open platform launches Cainiao agent, which can help users plan shipping options
- Alibaba Qwen App launches professional membership services, including office assistant and AI video generation packages
- Alibaba Qwen open platform goes live, with services such as renting homes and cars and sending express deliveries available via conversation