Microsoft Research Asia, an influential research institute based in Beijing and Shanghai, has continued to conduct AI research at high risk of abuse, including facial recognition, with sanctioned Chinese universities linked to the communist country’s military. Despite these ties Microsoft insists there are guardrails against such sensitive research at the institute.
Microsoft Research Asia (MSRA) was originally founded in 1998 as the Microsoft China Research Institute. By 2016, a Chinese media outlet reported that it had trained nearly 5,000 interns, with hundreds going on to Chinese tech giants and prestigious universities. It now employs over 300 scientists and researchers at its offices in Shanghai and Beijing.
As Microsoft’s position in China has come under increased scrutiny and U.S.-China relations have soured, Microsoft has tried to distance itself from MSRA’s deep connections with the communist country’s surveillance empire.
Microsoft claimed in 2024 that there are restrictions at MSRA “on work related to quantum computing, facial recognition and synthetic media” and a prohibition on “hiring or working with students and researchers from universities affiliated with China’s military.”
However, industry watchdogs claim that since March 2024, MSRA researchers have published ten papers that cover facial recognition or synthetic media which were coauthored with researchers from either sanctioned Chinese universities, military-linked institutions, or both. This suggests the “restrictions” on MSRA research into facial recognition and synthetic media are not particularly extensive and that such research continues despite the risks.
China experts tell Breitbart News that nine of these ten papers included coauthors from one of three sanctioned Chinese universities, two of which — Beihang University and the Harbin Institute of Technology — are considered part of the “Seven Sons of National Defense.” While not officially tied to the People’s Liberation Army, this unofficial group of universities is known for their very close connections to it.
The three sanctioned universities MSRA researched collaborated with on papers include:
- Beihang University, sanctioned by the US in March 2023 for its “contributions to China’s military modernization efforts.”
- Harbin Institute of Technology (HIT), sanctioned in 2020 “for acquiring and attempting to acquire U.S.-origin items in support of programs for the People’s Liberation Army,” particularly missile systems.
- The University of Science and Technology of China (USTC), sanctioned in May 2024 for supporting the advancement of China’s quantum technology development and China’s nuclear program.
MSRA quietly stopped recruiting from the “Seven Sons” and other universities with strong government ties, according to a 2022 report by QbitAI, a Chinese tech media outlet. But the research has continued, as evidenced by MSRA’s continued output of papers.
One of the papers, “PersonMAE: Person Re-Identification Pre-Training With Masked AutoEncoders,” embodies the risks inherent in much of the research MSRA does. The paper focuses on person re-identification, a subfield of computer vision concerned with tracking individuals across different cameras and locations.
This technology has clear applications in mass surveillance. A 2022 paper published in the Chinese journal Scientia Sinica Informationis, which also featured an MSRA researcher as a coauthor, noted that “as a key link in intelligent video surveillance systems, [person ReID] plays an important role in maintaining social public order,” likely referring to China’s vast network of real-time public surveillance cameras.
MSRA’s collaboration with universities linked to the People’s Liberation Army is not new. In 2019, the Financial Times revealed that Microsoft had been working with a university run by the Chinese Ministry of Defense on AI research, including facial recognition.
These types of ties are consistent with Microsoft’s attempt to position itself as anti-China while still profiting off of the Chinese Communist Party. As Breitbart previously reported, after Trump’s election, Microsoft asked for government “partnership” to challenge Chinese AI, while previously having invested over a $1 billion in AI research in China. Microsoft also claimed to not allow China to use ChatGPT, but made it still available via its Azure service.
Lucas Nolan is a reporter for Breitbart News covering issues of free speech and online censorship.
Read the full article here