Here’s how China has made mass surveillance a “killer application” for AI: will the US do the same?

It’s increasingly evident that the coronavirus pandemic will cause a radical re-shaping of many aspects of society, not least in the world of privacy. Many people are trying to discern the shape of that new world in the current evolving situation. An interesting analysis from Naomi Klein picks up on many themes that have appeared on this blog:

It’s a future in which our homes are never again exclusively personal spaces but are also, via high-speed digital connectivity, our schools, our doctor’s offices, our gyms, and, if determined by the state, our jails. Of course, for many of us, those same homes were already turning into our never-off workplaces and our primary entertainment venues before the pandemic, and surveillance incarceration “in the community” was already booming.

The whole piece is well-worth reading, if a little bleak. But it also serves the valuable purpose of pointing out a fascinating document obtained by the Electronic Privacy Information Center (EPIC) using US freedom of information laws. EPIC has been engaged in a long-running fight to obtain key material from the US National Security Commission on Artificial Intelligence (NSCAI). One of the many documents it received is a 2019 presentation entitled “Chinese Tech Landscape Overview“.

The NSCAI presentation is a good introduction to this increasingly important area – particularly for privacy – and includes a major section headed “AI/surveillance/Government as an Anchor Customer”. Its main point is that in China, “Mass surveillance is a killer application for deep learning”, and that “an entire generation of Al unicorns is collecting the bulk of their early revenue from government security contracts.” For example, AI companies such as Yitu and Sensetime advertise on their Web sites that Chinese police departments are using their facial recognition technology to catch criminals and solve murder cases. Similarly, the voice-recognition company iFlyTek is helping the Chinese police to eavesdrop on phone calls and obtain convictions as a result.

The presentation makes the point that wide-ranging support from the Chinese government means that AI startups can justify the high initial investment costs of major developments, confident they can recoup them by selling to the authorities, and in other sectors, since the technology also has applications outside surveillance. This gives them a key competitive advantage compared to AI companies in other countries.

As well as direct financial support in the form of part-ownership and future sales, AI companies benefit from the fruits of unrestrained government surveillance in China. Indigenous companies can draw on huge personal datasets to train and hone their products. The document estimates that the Chinese government has a database with over a billion faces of everyone older than 18 with an official ID, and that the police have a DNA database with 40 million samples. Although the databases have been created for surveillance, there are other applications that can be built on top of them.

For example, the presentation notes that the Chinese e-commerce giant Alibaba has been selected for the “National AI Team” in the area of “smart” cities. The massive street-side surveillance systems installed throughout China provide key data for optimizing urban life. As part of Alibaba’s “City Brain” project, the timing of red lights is adjusted for better traffic flow and to allow ambulances quicker passage through the streets during emergencies. It is reported that traffic time has been cut by 15.3%, and ambulance arrival time by 50% in test areas. The idea is to extend this to allow municipalities to make infrastructure decisions in real time, based on how every person is moving through the city.

As well as being able to make early investments secure in the knowledge that the Chinese government will provide contracts to help provide payback and allow access to uniquely extensive databases of personal information, there is another way in which AI companies benefit from the state’s activities. Among its other international trade programs, the Chinese government is keen to export its AI technology – notably for surveillance:

Chinese Al is already crossing borders – Cloudwalk [CloudWalk Technology] is helping to build a national facial database in Zimbabwe, and Yitu has begun selling image recognition tech to the Malaysian police; Kuala Lumpur. Malaysia’s capital. is integrating with Alibaba’s City Brain product.

Being the unambiguous world leader in Al would broaden China’s sphere of influence amongst an international community that increasingly looks to the pragmatic authoritarianism of China and Singapore as an alternative to Western liberal democracy.

The big question, according to the NSCAI presentation, is whether China’s lax approach to “data collection and privacy regulation” will provide it with unique advantages that allow it to establish itself as the world leader in AI. There’s a clear hint that the NSCAI would like to see a similarly lax approach in the US and elsewhere to counter that threat. In the agency’s interim report of November last year, it wrote: “GDPR could prove to be a significant obstacle in any efforts to standardize privacy regulations, which would be a key part of any international data sharing regime.” The NSCAI evidently thinks the GDPR provides too much privacy protection.

As Privacy News Online has reported, the coronavirus has already raised important issues about whether privacy protections should be relaxed in order to tackle the pandemic crisis. The worry has to be that Klein is right, and that digital giants will push for greater freedom to use personal data in the name of tackling the coronavirus – and boosting their profits along the way. One area where that is highly likely to be an issue is AI. Companies can probably expect to receive support from US government agencies like the NSCAI, which is clearly pushing to relax privacy protections in the US and globally.

Featured image by Gortu.