Rising Through the AI Ranks: Sensetime and the Potential for Biometric Surveillance

Sensetime, a Chinese company specializing in Artificial Intelligence (AI) applications across a variety of market segments, has been enlisted in the National Open Innovation Platform for Next-Generation AI alongside other prominent companies, which include Alibaba, Baidu, Tencent, and iFlyTek. The company has amassed a staggering investment total of US$4.5 billion, possibly making it the most prominent AI-based startup to date. The company’s vision is a multifaceted one, aimed at various market segments like automotive driver monitoring systems, augmented reality rendering for mobile applications, identity verification for banking and finance, and, of course, facial recognition software for surveillance initiatives. It is that latter application in which the Chinese government is highly interested, and it is one that is not only defining public security in modern China but also creating adjacency effects (i.e., biometrics surveillance) in Asia-Pacific (APAC) as a whole.

Amidst the maelstrom of AI and Machine Learning (ML) applications across the world, biometrics is one of the highly complementary technologies that can greatly benefit from the aforementioned innovative developments. However, putting AI to a true test would not include using it in a simple application to secure biometric authentication processes for a payment app or in a device verification as a part of a greater Multifactor Authentication (MFA) umbrella. Rather, there are much larger fish to fry. The convergence between AI technologies, biometrics, machine vision, edge computing, and advanced video analytics makes the realization of biometrics-based surveillance a truly fascinating—albeit controversial—future-looking application.

Biometricsurveillance refers to the use of biometric-based technologies as a means to gather information about individuals, to monitor their movements, and to identify them based on physiological (e.g., fingerprint, iris, face) and behavioral (e.g., voice, behavior, gait) characteristics.While law enforcement, military, border control, transportation, and public security will be the primary adopters of biometric surveillance initiatives in the near future, there are other highly important applications that will emerge from social media, automotive vehicles, smart homes, and smart cities. Thus, the coexistence of well-honed and responsively trained AI algorithms, facial recognition software, surveillance, and advanced analytics capable of analyzing (and even predicting) human behavior not only is a “happy accident” but also is a highly anticipated technological outcome. This technology is needed in order for biometrics to escape the confines of its mathematically probabilistic nature and ascend to its true potential as a more accurate technology. As such, the likelihood that biometrics will be inherently more vulnerable to inaccurate measurements (which may generate a fair amount of criticism, especially regarding law enforcement applications) will be significantly reduced, and the path will be open for greater acceptance by more vendors and governments worldwide.

But what will an AI-fueled biometric surveillance entail? And, perhaps more importantly, will the progress from related applications be justified or generate even more problems than solutions?

ABI Research posits that “sharpening” the accuracy and training of AI algorithms will have a direct effect on the entire biometrics industry as whole. This will extend into a wide spectrum, including everything from consumer (simple authentication to online payments), commercial, and enterprise (MFA, access control, workforce management) applications, all the way to military and law enforcement uses. However, it is the advances in biometric surveillance that will not only close the security gap in certain high-profile markets (e.g., border control, transportation) but also extend the reach of law enforcement and even compel an unprecedented and dynamic shift into how modern societies approach the concept of public security.

In the case of the megacities across the world (e.g., Shanghai, Tokyo, Delhi, New York, Cairo, etc.), biometric surveillance will become a vital part of future dragnet operations. The constant threat of terrorism will force all law enforcement agencies to turn to biometrics; this should prove to be an effective (albeit still questionable) technology for mass surveillance operations. This effect will slowly extend to reach other Internet of Things (IoT) verticals such as connected vehicles, smart cities, and smart homes. In essence, every vertical that can provide biometric data of any range, modality, or accuracy can and will be monitored by authorities and law enforcement, ranging from public-facing applications all the way to consumer electronics. This will differ across geographic locations, and fewer legislative and regulatory restraints will be placed for countries in the developing world or emerging economies. ABI Research predicts that even geolocations such as the European Union that have enforced restrictions and data protection laws (e.g., the General Data Protection Regulation) will amend the legislation to allow for law enforcement and surveillance to retain users’ and citizens’ biometric data for a much longer time period.

It is expected that the backlash from citizen organizations will be unrelenting and perfectly justified—especially in countries with a more democratically based regime. The potential for corruption and unlawful application is and will always be a factor in biometric applications, and, if history has taught security circles anything, it is that this has always been the case with almost any new technology or application that could reach millions, if not billions, of users. Just a brief glance into Edward Snowden’s leaked documents shows us only a small picture of what is currently being applied in the background as well as in plain sight. However, progress cannot be easily stopped, and companies operating under aforementioned verticals are advised to keep an open eye in the near future. This relates both to the potential to expand their offerings toward government, military, and law enforcement, while also taking into account the constantly changing regulations about biometric data storage and management. This shift is already underway in APAC and particularly in China, where surveillance and facial recognition capabilities are expected to enjoy a considerable surge in the next five to 10 years.