MBZUAI Partners With US Tech, Collaborates With PLA Scientists
Publication: China Brief Volume: 24 Issue: 17
By:
Editor’s Note: This article is the second in a two-part series. The first part can be found here. In response to the previous article, The Jamestown Foundation received a letter from Eric Xing, President of the Mohamed bin Zayed University of Artificial Intelligence. The letter can be read in full here, or by going to that previous article and scrolling to the bottom.
Executive Summary:
- There are connections between the Mohamed bin Zayed University of Artificial Intelligence in Abu Dhabi (MBZUAI) and People’s Liberation Army (PLA) and PLA-affiliated institutions. This includes research collaborations with scientists from those institutions, as well as a career pipeline between Abu Dhabi and the People’s Republic of China (PRC).
- Research collaborations funded by the PRC have produced articles published in top journals on topics that have military applications, such as machine learning algorithms, network security, and Internet of Things technologies.
- MBZUAI has access to technologies that are currently subject to US export controls, such as NVIDIA A100 graphics processing units (GPUs). It is plausible that these are being used in research that assists in the development of PLA capabilities.
The Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), a research university in Abu Dhabi, has ties to the united front system in the People’s Republic of China (PRC). It also has access to US technologies that are currently subject to US export controls, such as NVIDIA A100 graphics processing units (GPUs). This raises the possibility that the university could function as a site for entities or individuals from the PRC to acquire or gain access to such technologies, thereby circumventing US export controls. Numerous PRC researchers are based at MBZUAI, many of whom come from People’s Liberation Army (PLA) or PLA-affiliated institutions. Meanwhile, US companies such as IBM and Meta, and academic institutions such as the University of Michigan and University of California, Berkeley, have partnered with MBZUAI (China Brief, August 15). The PRC’s weaponization of international cooperation to bolster its defense capabilities suggests the potential scale of this problem. Ultimately, research done at MBZUAI and using US technology could later be used to enhance the PRC’s military capabilities.
International Collaboration to Boost Defense Modernization
The PRC has long advocated for international technology cooperation. In November 2023, at the inaugural “One Belt One Road” Technology Exchange Conference (“一带一路” 科技交流大会), the PRC proposed an “International Science and Technology Cooperation Initiative (国际科技合作倡议).” This initiative advocates for building “a global community of science and technology (全球科技共同体)” and an “open and free ecosystem of international technology cooperation.” The aspiration is for “scientific and technological innovation personnel and resources to flow freely around the world” (Ministry of Science and Technology [MOST], November 7, 2023).
This initiative appears to emphasize global collaboration and openness. However, as CCP Chairman Xi Jinping stated at a meeting in June 2024, part of this openness is intended to buttress the PRC’s military capabilities. In a speech delivered at an event in Beijing where he presented the country’s most prestigious science and technology award, he argued that “to build a technologically strong country, [the PRC] must possess world-leading technology and innovation capabilities to support leaps in economic strength, defense capabilities, and comprehensive national power” (Xinhua, June 24).
International cooperation remains one of the most efficient ways for the PRC to access advanced technologies that are critical to its national defense. In 2018, a professor at the PLA’s National University of Defense Technology (NUDT; 国防科技大学) concluded an article in the PLA Daily newspaper by writing that if the PRC “deepens international exchanges and cooperation and fully uses global innovation resources,” it will “advance independent innovation in national defense technology from a higher starting point” (PLA Daily, March 28, 2018).
MBZUAI Hires Scientists From PLA-Affiliated Universities
MBZUAI has recruited scientists with connections to the PLA and the PRC’s defense industry. One professor at MBZUAI’s Machine Learning Department, Mohsen Guizani, has a long history of research collaboration with PLA scientists. He has published 17 research papers with Fu Xiao (伏晓), an associate professor at the Software Institute at Nanjing University who has received funding for multiple projects from the PLA (Nanjing University, accessed August 21). Gao Jianbin (高建彬), who has hosted research projects for the PLA’s Information Technology Safety Research Center, is another collaborator. Guizani has also worked on various projects with researchers from the PLA University of Science and Technology (Jiangsu Internet of Things Technology and Application Collaborative Innovation Center, August 22, 2018; Free Kaoyan, September 12, 2021). In total, Guizani has co-authored at least 167 articles in journals published by the Institute of Electrical and Electronics Engineers (IEEE) alongside PRC-affiliated researchers. Thirty-one of these papers involved collaboration with individuals from PLA-affiliated universities. [1]
Xiong Huan (熊欢), an assistant professor of machine learning at MBZUAI, is also a professor at the Harbin Institute of Technology (HIT; 哈尔滨工业大学). His bio on MBZUAI’s website, however, makes no reference to his academic career in the PRC (HIT, December 20, 2023; arXiv, Jul 30; MBZUAI, Accessed August 20). Gu Bin (顾彬), another assistant professor of machine learning, received his PhD from the Nanjing University of Aeronautics and Astronautics. Both of these institutions are among the “Seven Sons of National Defense (国防七子)” (See Appendix 1.). [2] At least four other MBZUAI researchers and professors have received degrees from other “Seven Sons” universities. [3]
MBZUAI has partnered with institutions that have been blacklisted by the US government for their ties to the PLA and has employed scientists from such institutions. This is perhaps best exemplified by an October 2019 visit by MBZUAI’s then-CEO Shao Ling (邵岭) to the Beijing Institute of Technology (BIT; 北京理工大学), another “Seven Sons” university. While there, he hosted an information session on MBZUAI’s scholarship program, actively seeking to attract BIT students to the university (BIT, October 14, 2019).
This relationship also operates in the opposite direction, where researchers who work at MBZUAI later move to “Seven Sons” universities. For instance, Guan Dayan (官大衍), now an associate professor at HIT, previously served as a research scientist at MBZUAI from March 2022 to July 2023 (HIT, July 12). Similarly, Xie Guo-Sen (谢国森), who was a researcher at MBZUAI from 2020 to 2022, is now a professor at the Nanjing University of Science and Technology (NJUST; 南京理工大学) (NJUST, Accessed June 13).
MBZUAI Research Collaborations with PLA Scientists
MBZUAI has collaborated with scientists at each of the “Seven Sons” universities. It has not, however, only collaborated with institutions that have close ties to the military establishment. It has also worked directly with scientists and researchers at PLA universities. These include the PLA Information Engineering University (PLAIEU; 中国人民解放军战略支援部队 信息工程大学) (formerly under the now-defunct Strategic Support Force), NUDT, the PLA Army Engineering University (AEU; 中国人民解放军陆军工程大学), and the Academy of Military Sciences. The focus of these research projects, which have taken funding from PRC institutions, have often aligned with goals that have been articulated in PRC strategy documents, focusing on areas such as communications and network security, cloud computing and Internet of Things (IoT) security, and cybersecurity more generally. This research could also be applied for offensive purposes (ASPI, November 25, 2019). NUDT and PLAIEU are directly engaged in cyber operations (ASPI, May 13, 2017).
A paper published in the journal IEEE Transactions on Dependable and Secure Computing as a result of one such research collaboration details a new protocol to make messaging over 5G networks more secure and more efficient (IEEE Xplore, December 9, 2022). Another paper, this time co-authored by Guizani and researchers at NUDT and the Academy of Military Sciences and published in the journal IEEE Transactions on Mobile Computing, focuses on IoT research. The paper proposes a new method of detecting problems in IoT systems and making them more secure (IEEE Xplore, November 6, 2023).
Much more research stemming from collaborations between MBZUAI, NUDT, and others has been published since the gulf state university was founded five years ago. A collaboration between MBZUAI, NUDT, and others, including the University of Kent in the United Kingdom, looked at “digital twin” technology, using virtual simulations of systems to study and optimize them. Specifically, the research investigated improving the performance and efficiency of Vehicle-to-X (V2X) communication in 6G networks (IEEE Xplore, October 24, 2023). Another research collaboration between MBZUAI, NUDT, and JD Explore Academy proposed a new machine learning algorithm to optimize performance in certain data processing scenarios (arXiv, October 8, 2023).
MBZUAI researchers again teamed up with AEU, this time with others, including the University of Houston, for a separate project on “digital twin” technology. In a paper they published, the researchers proposed a framework for digital twin-enabled wireless systems for better protection against cyberattacks (IEEE Xplore, November 20, 2023). Other MBZUAI scientists have also published research with academics at the Academy of Military Sciences and other PRC universities on innovative approaches to network security (IEEE Xplore, June 26, 2023).
MBZUAI has also collaborated extensively with military-affiliated institutions in the PRC, including “Seven Sons” universities. Researchers at MBZUAI and Beihang University have presented papers at NeurIPS, one of the most prestigious machine learning conferences, for example. Beihang University specializes in aerospace and information technology (arXiv, October 7, 2022; BUAA, accessed August 29). This work was partially done while Huang Lei (黄雷) was a visiting scholar at MBZUAI. Huang is currently an associate professor at Beihang University’s Artificial Intelligence Institute, where he received his PhD. He was also a research scientist at the Inception Institute of Artificial Intelligence, another institution in the United Arab Emirates with close PRC ties (Beihang University, March 22, 2021). Other projects include work on machine learning optimization problems led by HIT PhD student Xinzhe Yuan (袁新哲) with contributions from MBZUAI professors Xiong Huan and Gu Bin, and work with Harbin Engineering University on IoT security for connected vehicles and other applications (IEEE Xplore, December 23, 2021; IEEE Xplore, June 20, 2023; MBZUAI, May 23). Zheng Peng (郑鹏), who co-authored research on computer vision techniques, was affiliated with MBZUAI while simultaneously a graduate student at the Nanjing University of Aeronautics and Astronautics (NUAA; 南京航空航天大学) (Association for the Advancement of Artificial Intelligence, June 25, 2023). During his time as a research assistant at MBZUAI, he was supervised by an NUAA Professor (Peng Zheng, March 23; NUAA, accessed August 29). Finally, MBZUAI scientists have collaborated with academics at NJUST, also on computer vision (arXiv, March 26, 2021).
All these research projects are upstream of potential military applications. For instance, as the PRC focuses more on cyber warfare and informatized warfare, the security and efficiency of networks has become increasingly important. Similarly, using machine learning techniques such as computer vision to enhance object identification in surveillance imagery can improve the accuracy and effectiveness of threat detection, targeting, and situational awareness. As IoT technologies progress, connected vehicles and other networked devices are changing how modern militaries can operate. Much of the research on which MBZUAI and PRC scientists have collaborated focuses on more theoretical approaches to solving narrowly defined problems. These foundational innovations form the basis for military technologies that other PLA scientists likely are seeking to develop and deploy.
Conclusion
MBZUAI has numerous ties with PLA institutions, as well as with the universities known as the “Seven Sons of National Defense.” These include research collaborations with scientists at all seven of those universities, as well as a career pipeline, with a number of researchers moving between MBZUAI and the “Seven Sons.” This presents challenges to the US government’s attempts to restrict technology transfer to the PRC over fears that US technology could be used to support the development of the PRC’s military capabilities. NUDT and the “Seven Sons” appear on the US Department of Commerce Entity List (Code of Federal Regulations, Accessed September 3). Yet MBZUAI provides a potential pathway for these entities to bypass US restrictions, as it has access to technologies such as NVIDIA A100 GPUs. This suggests that more rigorous research into both corporate and academic partnerships must be done if the US government is to fix the problems its export controls are intended to solve.
Notes
[1] IEEE is a global professional organization known for its central role in advancing technology and innovation. It publishes some of the most influential academic journals in the fields of electrical engineering, electronics, and computer science, and hosts numerous conferences each year (IEEE, accessed September 3). In this article, the term “PRC-affiliated researchers” refers to individuals who have affiliations with universities or institutions within the PRC as noted in the IEEE publications. PLA-affiliated universities include both universities with direct PLA links and the “Seven Sons of National Defense” universities (see [2]).
[2] The “Seven Sons of National Defense” are a consortium of universities directly overseen by the PRC’s defense industrial authority which are known for their close ties to the military establishment and which have been blacklisted by the US government (ASPI, November 25, 2019; eCFR, accessed August 31).
[3] These are: Xu Min (徐旻), MBZUAI Affiliated Associate Professor of Computer Vision, who got his Bachelor of Engineering in computer science from Beihang University (MBZUAI, accessed August 20; Xu Lab, accessed September 3); Gao Zhengqing, who became a PhD student in machine learning at MBZUAI after graduating from HIT (LinkedIn, accessed August 20); Guokan Shang (尚国侃), who is a Lead Research Scientist at MBZUAI France Lab and received a Bachelor’s degree in 2013 from HIT (Guokan Shang, accessed August 21); and Liu Nian (刘念), a research scientist at MBZUAI who received his PhD from the Northwestern Polytechnical University (Google Sites/liunian228, accessed August 29).
Appendix 1 Examples of MBZUAI’s collaboration with the ‘Seven Sons of National Defense’
Defense universities | Research projects | PRC Funding | Other institutions involved | Link to Published Research |
Beihang University | Adaptive Optics Compensation for Orbital Angular Momentum Optical Wireless Communications | Yes | Beijing Institute of Technology; Beijing University of Posts and Telecommunications | (IEEE Xplore, July 19, 2022) |
Understanding Whitening Loss in Self-supervised Learning | Yes | University of California, Merced (USA); Google | (IEEE Xplore, June 21) | |
An Investigation into Whitening Loss for Self-supervised Learning | Yes | N/A | (arXiv, October 7, 2022) | |
Active Aerial Reconfigurable Intelligent Surface Assisted Secure Communications: Integrating Sensing and Positioning | Yes | Northwestern Polytechnical University; Hosei University (Japan); Xi’an Jiaotong University; Shenzhen MSU-BIT University (a PRC, Russia joint university); University of British Columbia (Canada) | (IEEE Xplore, June 14) | |
An Optimized Plane Detection-Based Topological Metric for Indoor LiDAR SLAM | Not disclosed | Wisedawn Auto Co., LTD | (SSRN, January 10) | |
Achieving Fast Environment Adaptation of DRL-Based Computation Offloading in Mobile Edge Computing | Yes | Institute of Software Chinese Academy of Sciences | (IEEE Xplore, September 28, 2023) | |
Management of Positioning Functions in Cellular Networks for Time-Sensitive Transportation Applications | Yes | Nanjing University of Aeronautics and Astronautics; Beijing Institute of Technology; University of Edinburgh (UK) | (IEEE Xplore, January 20, 2023) | |
Beijing Institute of Technology | Multi-Level Representation Learning With Semantic Alignment for Referring Video Object Segmentation | Not disclosed | Inception Institute of Artificial Intelligence; Terminus Group; University of Macau | (IEEE Xplore, September 27, 2022) |
Harbin Institute of Technology | Special Characters Usage and Its Effect on Password Security | Yes | Jiangxi University of Science and Technology; City University of Hong Kong | (IEEE Xplore, February 19) |
A Malicious Domains Detection Method Based on File Sandbox Traffic | Yes | East China Normal University; City University of Hong Kong | (IEEE Xplore, October 25, 2022) | |
A Comprehensive Detection Method for the Lateral Movement Stage of APT Attacks | Yes | East China Normal University; State Key Laboratory of Public Big Data, Guizhou University; City University of Hong Kong | (IEEE Xplore, October 6, 2023) | |
Dynamic Spiking Framework for Graph Neural Networks | No | Hebei University of Technology; Microsoft Corporation; Technology Innovation Institute (TII, UAE); Jilin University | (arXiv, July 30) | |
NDOT: Neuronal Dynamics-based Online Training for Spiking Neural Networks | No | TII; Sant’Anna
School of Advanced Studies (Italy); Jilin University
|
(OpenReview, accessed August 20) | |
New Insight of Variance reduce in Zero-Order Hard-Thresholding: Mitigating Gradient Error and Expansivity Contradictions | Not disclosed | Jilin University | (MBZUAI, May 23) | |
Harbin Engineering University | AI-based Intrusion Detection for Intelligence Internet of Vehicles | Yes | N/A | (IEEE Xplore, December 23, 2021) |
Privacy-Preserving Outsourcing of K-Means Clustering for Cloud-Device Collaborative Computing in Space-Air-Ground Integrated IoT | Yes | N/A | (IEEE Xplore, June 20, 2023) | |
Northwestern Polytechnical University | GP-NeRF: Generalized Perception NeRF for Context-Aware 3D Scene Understanding | Yes | Hefei University of Technology; Nanyang Technological University (Singapore); Baidu Inc.; Hefei Comprehensive National Science Center | (Computer Vision Foundation, accessed August 20) |
Learning Non-Target Knowledge for Few-Shot Semantic Segmentation | Yes | IIAI; National Center for Artificial Intelligence of the Saudi Data and AI Authority | (IEEE Xplore, 2022; IEEE Xplore, September 27, 2022) | |
Nanjing University of Aeronautics and Astronautics | Multi-UAV-Assisted Federated Learning for Energy-Aware Distributed Edge Training | Yes | Guizhou University; Nanyang Technological University (Singapore); Singapore University of Technology and Design | (IEEE Xplore, July 24, 2023) |
Memory-Aided Contrastive Consensus Learning for Co-salient Object Detection | Yes | ETH Zurich; IIAI | (Association for the Advancement of Artificial Intelligence, June 25, 2023) | |
GCoNet+: A Stronger Group Collaborative Co-Salient Object Detector | Yes | Institute of High Performance Computing (Singapore); ETH Zurich; Hong Kong University of Science and Technology | (MBZUAI, September 1, 2023) | |
Nanjing University of Science and Technology | Non-Salient Region Object Mining for Weakly Supervised Semantic Segmentation | Yes | University of Electronic Science and Technology of China; University of Adelaide; University of Technology Sydney | (arXiv, March 26, 2021) |