By Alex Joske
For more than a year, debate has raged over allegations that the Chinese military is taking advantage of Google’s research and expansion into China. General Joseph Dunford, chairman of the US Joint Chiefs of Staff, told a senate committee in March that Google’s work in China indirectly benefits the Chinese military, an accusation echoed by President Donald Trump. Google’s response was unequivocal: ‘We are not working with the Chinese military.’
There is no evidence that Google has a direct relationship with the People’s Liberation Army, but, as with the collaboration seen between many Western universities and the PLA, artificial intelligence researchers from the Chinese military have worked with Western technology firms’ employees on research that could advance China’s military capability.
The situation reflects the lack of a clear policy on engagement with Chinese entities across universities, companies and governments.
A scientist from the PLA and a Google employee were among the five co-authors of a paper related to artificial intelligence published in January 2019. The lead author, Guan Naiyang, is an associate professor at the PLA National University of Defense Technology (NUDT). His research focuses on non-negative matrix factorisation, an algorithm used in machine learning.
Guan’s story is emblematic of the PLA’s efforts to leverage overseas expertise. He received all his university degrees from NUDT but worked abroad as a visiting PhD scholar. In 2012, he visited the University of Technology Sydney, studying under one of Australia’s leading AI researchers, who is also a visiting professor at NUDT. His doctoral thesis earned him top prizes from the PLA and the China Computer Federation. Australian law doesn’t yet regulate the transfer of technology and training to foreign nationals or members of foreign militaries who are physically in Australia.
Guan worked on three PLA projects about online-surveillance and intelligence-collection technology when he was a PhD student. From what little information is available, we can establish that at least one of the projects—focused on analysing intelligence sourced from the internet—relied on the same kind of technology discussed in the paper he co-authored with a Google employee.
Last year, two NUDT scientists worked at a lab at Princeton University as visiting scholars. Their supervisor and colleague at Princeton is also a senior staff research scientist at Google and worked with the NUDT scientists on computer vision. Other papers written by the NUDT scientists examined target detection in sea clutter and the automatic recognitionof objects including planes.
While most scientists in academia dedicate themselves to expanding our knowledge of the world, Guan’s goal, and the goal of PLA scientific research in general, is different. In 2016, he told a PLA newspaper: ‘I want to hasten the software development and application of high-performance computers, comprehensively propelling artificial intelligence toward the battlefield.’
Scientists like Guan and those who visited Princeton are among the thousands of PLA officers and cadres who have been sent abroad as PhD students or visiting scholars in the past decade. In Picking flowers, making honey, an ASPI report published last October, I analysed these activities in detail and showed how the Chinese military exploits the openness of academic institutions to improve its own technology and expertise. The report’s title comes from a saying the PLA has used to describe its international collaboration: ‘Picking flowers in foreign lands to make honey in China.’
Many Western companies and their employees have worked with the Chinese military in ways that could advance its intelligence and warfighting capabilities. A Financial Timesarticle recently uncovered Microsoft’s ties to Chinese military AI researchers. Since at least 2010, Microsoft’s Asian research arm has taken interns from the PLA.
It shouldn’t be much of a surprise that companies such as Google and Microsoft have been caught up in the PLA’s efforts to leverage domestic and overseas expertise. Universities often engage in little scrutiny of their Chinese partners; leading universities in Germany, Australia, Norway, the US and the UK have all accepted Chinese military officers who claimed to be from non-existent institutions as visiting scholars. Some companies and even governments have made similar mistakes.
It’s encouraging to see that efforts are emerging to develop clearer policy guidance and regulation to help universities and companies understand and address this critical national security problem, although much more needs to be done.
Collaboration with the PLA often crosses a red line, but activities that indirectly benefit the Chinese military pose a tough challenge. Military–civil fusion, the Chinese government policy that’s pushing the PLA to cultivate international research ties, is also building greater integration between Chinese civilian universities and the military. As ASPI non-resident fellow Elsa Kania has pointed out, Google’s work with Tsinghua University is worrying because of the university’s growing integration with the PLA.
This raises a troubling question: if a company, government or university is unable to control collaboration with overt Chinese military entities, how can it effectively manage more difficult areas, like collaboration with military-linked entities?
Without clear policies and internal oversight, Western tech firms that don’t intend to work with the PLA may have employees who are doing so. Greater debate and robust policies are needed to ensure that universities and companies avoid contributing to the Chinese military and technology-enabled authoritarianism, and don’t inadvertently give fuel to those wanting an end to all collaboration.
A starting point here is for governments to begin setting out clear policy guidance and improving export controls to target entities that universities and companies should not be collaborating with. In the meantime, self-regulation and internal oversight by these companies will help address government concerns—and help inform future regulation. Civil society—NGOs and media—can also develop resources to help universities improve their engagement with China. ASPI’s International Cyber Policy Centre is currently developing a database of Chinese military and military-linked institutions for this purpose.
This article was first published by ASPI on April 12, 2019.
Alex Joske is a researcher at ASPI’s International Cyber Policy Centre.
For a look at the impact of such efforts on the Soviet Union, see the following: