Written By Angshuman Kaushik
Mr Kaushik is an Ex-Judge turned to an AI Safety Researcher
The rapid expansion of Facial Recognition Technology (FRT) by law enforcement agencies globally raises significant concerns about fundamental rights. While offering security benefits, FRT's unchecked use risks discrimination, privacy violations, and political misuse, especially in an era where policy and public understanding lag behind technological advancements.
BACKGROUND
The use of Facial Recognition Technology (FRT) by law enforcement agencies in public places is rapidly expanding. Both democratic and authoritarian governments are increasing its deployment within civil society. Several factors contribute to this growth. Firstly, the rising incidence of criminal activities affects diverse societies globally, irrespective of their political ideologies or economic standards. As advancements in communication, weaponry, and data technologies enhance criminal capabilities and reach, law enforcement agencies face mounting pressure to counteract these developments.
Secondly, law enforcement agencies enforce a variety of civil control agendas across societies. These include targeted control of specific populations, such as racial and ethnic minorities, genders, age-defined groups like children and the elderly, and health-defined populations such as individuals with disabilities. This control extends to broader social rights areas, including the right to respect for private life, the protection of personal data, freedom of assembly and association, freedom of expression, and the right to an effective remedy and a fair trial. Fundamental rights concerns also arise from the vulnerable position of individuals whose facial images are captured and processed.
Thirdly, law enforcement agencies are often employed to enforce specific political agendas. The use of FRT as a tool for political control by authoritarian regimes is well-documented. Less commonly recognized is the similar practice in more democratic societies. As recent authoritarian tendencies emerge within democracies, the use of FRT technology by law enforcement becomes increasingly concerning.
Although FRT offers security benefits to the public, its unchecked use can have a detrimental effect on the fundamental rights of citizens. It is essential to examine a few of these concerns to further define the problem.
One frequently raised concern about FRT is the risk of errors in matching faces. A study by the National Institute of Standards and Technology (NIST) found that FRT has higher error rates when used on people of color, women, the elderly, and children. This suggests that the technology can discriminate based on factors like race and gender. Moreover, while facial recognition technologies perform well under consistent conditions—such as when all photographs are taken with similar lighting and from a frontal perspective—significant errors can occur when photographs involve varying lighting, poses, and backgrounds.
From a U.S. perspective, a recent report titled "American Dragnet" reveals that Immigration and Customs Enforcement (ICE) has used facial recognition technology to search through the driver’s license photographs of about 1 in 3 (32%) of all adults in the U.S. The report further states that despite the vast scope and clear civil rights implications of ICE’s surveillance practices, the agency has managed to keep these practices shrouded in near-total secrecy, avoiding enforcement of even the few laws and policies that could impose limitations.
At the federal level, the U.S. Government Accountability Office (GAO) recently surveyed 42 federal agencies employing law enforcement officers about their use of FRT. Twenty agencies reported either owning facial recognition systems or using systems owned by others, including federal, state, local, and non-government entities. Besides legal and policy issues, the use of FRT by U.S. law enforcement, in combination with photographic or video surveillance, may also raise considerations under the U.S. Constitution.
When combined with other technologies and practices, FRT poses additional risks. In recent years, the capabilities of surveillance cameras have significantly increased due to the addition of sophisticated video analytics. For example, in 2010, less than 2% of network cameras sold featured embedded video analytics, but this proportion grew to over 40% by 2016 and continues to rise. These analytics increasingly rely on artificial intelligence, and the added capabilities to perform facial recognition and identify suspicious behavior are among the most problematic features of advanced video surveillance systems. In July 2020, the UN Human Rights Council adopted a resolution expressing concern over the unlawful or arbitrary surveillance of individuals engaged in peaceful protests through the use of facial recognition technologies.
CONCLUSION
There is limited information about the way and extent to which facial recognition technology is being used by law enforcement agencies and about the impact of its use upon fundamental rights. For instance, although there is more limited information available about the actual use of FRT in the European Union, several Member States are considering, testing, and implementing its use for law enforcement purposes. This lack of more comprehensive information about the actual use of the technology limits opportunities to analyze its fundamental rights implications.
Currently, the average “man-on-the-street” has little substantive, authoritative information regarding this new FRT technology. In addition, much of that is grounded in misinformation, conspiracy theories and other types of information detrimental to the smooth functioning of civil society. Both individuals and society in general would benefit from open, authoritative, and widely dispersed information.
Comments