Study examines the need to balance innovation and privacy in facial recognition

A recent report under the auspices of the Carr Center for Human Rights Policy at Harvard University, Artificial Intelligence: Promises, Risks, and Regulation, says the use of facial recognition technology (FRT) in surveillance raises significant ethical and legal concerns. The report underscores the potential dangers of mass surveillance, particularly in authoritarian regimes where governments can use AI-driven facial recognition to monitor and suppress political dissent.

One of the report’s authors, Luís Roberto Barroso, a Senior Fellow at Harvard Kennedy School, said, “Intuitively, the problem lies not with the technology itself, but in how we use it and, more importantly, in how we intend to distribute the benefits it will generate. Hence, the challenge ahead is to create an institutional design that promotes the beneficial uses of AI and limits its abuses, preventing the automated production of injustice and the multiplication of existing risks.”

Within this broader discussion, facial recognition emerges as one of the most controversial and impactful developments. As AI becomes more embedded in our daily lives, understanding how facial recognition works, its applications, risks, and potential regulatory frameworks is crucial to shaping its ethical use, the report cautions, noting that facial recognition technology relies on AI algorithms to identify or verify a person’s identity based on facial features. These systems analyze unique facial structures, matching them with existing databases to authenticate identities.

While FRT has been in development for decades, recent advancements in machine learning and neural networks have significantly improved its accuracy and efficiency. But immature and problematic AI can get things wrong. And yet, it’s now widely used in everything from law enforcement and border security to commercial applications such as unlocking smartphones and personalized advertising. Its integration into daily life suggests a future where physical identification documents may become obsolete, the report suggests.

One of the most prominent applications of facial recognition is in security and law enforcement. Governments worldwide have implemented FRT to enhance surveillance, prevent crime, and track suspects. Police departments utilize the technology to identify individuals in crowds, assisting in criminal investigations and counterterrorism operations. Airports and border control agencies employ facial recognition to streamline immigration procedures, reducing wait times and enhancing security. These uses demonstrate the efficiency and convenience of FRT in ensuring public safety.

The inherent problem in all this is the questionable ability to track individuals without their consent, infringing on fundamental human rights, including privacy and freedom of expression. In some countries, AI-powered surveillance has been linked to political persecution and discrimination, reinforcing concerns about the unchecked deployment of this technology.

Regarding privacy, the report says, “at least three aspects require attention.” One is the collection of Internet users’ data without their consent by digital platforms and websites. This information is used for com­mercial purposes, such as targeting information and advertising, or even for manipulating the user’s will, as neuroscience research has shown.”

Another aspect, the report says, “concerns surveillance and tracking by government and police authorities using facial recognition technologies and tracking tools. Although the objective is to fight crime, the risks of abuse are very high, especially in authoritarian governments.”

Finally, “AI systems require vast amounts of data to train their models, which introduce risks of data leakage and cyberattacks by malicious actors. For example, spear phishing and doxing can fuel harass­ment, political violence, malinformation, and disinformation.”

Beyond law enforcement, facial recognition is increasingly being adopted by private corporations. Retailers use FRT to analyze consumer behavior, offering personalized advertisements and improving customer service. Social media platforms integrate facial recognition to automate tagging in photos and enhance user engagement.

While these applications offer convenience and customization, the report warns that they also introduce significant risks regarding data security and user consent. The collection and storage of biometric data pose privacy threats, as breaches could expose individuals to identity theft and unauthorized surveillance.

Another pressing concern associated with facial recognition technology is bias and accuracy. The report points to studies that have shown that facial recognition algorithms often exhibit racial and gender biases, leading to higher error rates for individuals with darker skin tones and non-male gender identities.

These biases stem from training datasets that predominantly feature lighter-skinned individuals, resulting in discriminatory outcomes. The report underscores the risk of perpetuating systemic inequalities through AI, particularly in judicial and policing contexts where misidentifications can lead to wrongful arrests and convictions. Addressing these biases requires more diverse and representative data, as well as transparency in how algorithms are trained and deployed.

“Algorithms are trained with existing data, which inherently reflect past and present human behaviors imbued with bi­ases and prejudices shaped by historical, cultural, and so­cial circumstances,” the report says. “Consequently, they tend to perpetuate current and past social structures of inclusion and exclu­sion.”

Biometric testing by the National Institute of Standards and Technology (NIST) has found that the majority of facial recognition algorithms are more likely to misidentify people with darker skin, women and the elderly, though the most accurate algorithms show very low differentials in the Institute’s latest testing.

The report says that the regulation of facial recognition technology is an essential step in mitigating its risks while preserving its benefits. The report suggests that governments and international organizations must establish legal frameworks to govern the ethical use of FRT. Regulations should include strict guidelines on data collection, ensuring that individuals provide informed consent before their biometric data is used. Additionally, oversight mechanisms should be in place to prevent misuse and ensure accountability in both public and private sector applications.

Some countries have already taken steps toward regulating facial recognition technology. The European Union, through its General Data Protection Regulation (GDPR), imposes strict rules on biometric data processing, requiring explicit consent and justification for its use. In the United States, certain cities have banned the use of facial recognition by law enforcement due to concerns about privacy and civil liberties. However, a global consensus on AI regulation remains elusive, highlighting the need for coordinated efforts to establish ethical standards and safeguards.

Despite its challenges, the report says, facial recognition technology holds significant potential when used responsibly. In healthcare, FRT can aid in diagnosing genetic disorders by analyzing facial features associated with specific conditions. It can also assist visually impaired individuals by providing real-time identification of people and objects.

Additionally, in financial services, facial recognition enhances security by providing biometric authentication for transactions, reducing fraud and unauthorized access. These applications demonstrate that, when used ethically, FRT can contribute positively to society.

To balance innovation and ethical considerations, though, the report says policymakers, technologists, and civil society must collaborate in shaping the future of facial recognition technology.

The report stresses the importance of AI governance rooted in human rights principles, ensuring that technological advancements do not come at the expense of individual freedoms. By fostering transparency, accountability, and inclusivity in AI development, societies can harness the benefits of facial recognition while minimizing its potential harms.

“It seems that the regula­tion of AI is indispensable, however, the task is not simple and faces numerous challenges and complexities,” the report acknowledges, noting that “regulation needs to be done on a moving train. In March 2023, over a thousand scientists, researchers, and entrepre­neurs signed an open letter calling for a pause in the development of the most advanced AI systems given the ‘profound risks to society and humanity’ they represented. The pro­posed pause, for at least six months, aimed to introduce ‘a set of shared security protocols.’ While the concerns were fully justified, research was not suspended. The train continued at high speed, driven by the competitive race among nations, researchers, and entrepreneurs. Nevertheless, the letter reinforced the urgent need for governance, regulation, monitoring, and attention to the social, economic, and politi­cal impacts of new technologies.”

The report warns that “the speed of change is astonishing,” and that “makes it extreme­ly difficult to predict future developments and adapt legal norms accordingly, which risk becoming obsolete quickly.”

Ultimately, the future of facial recognition technology depends on how it is regulated and implemented, the report says. As AI continues to evolve, it is imperative to establish policies that protect human rights, address biases, and prevent the misuse of biometric data.

Public awareness and informed debate are crucial in determining the ethical boundaries of facial recognition, ensuring that its use aligns with democratic values and societal well-being. By proactively addressing these challenges, governments and institutions can create a framework that maximizes the benefits of facial recognition while safeguarding individual liberties and promoting ethical AI deployment.

Article Topics

biometric identification  |  biometrics  |  ethics  |  facial recognition  |  regulation  |  responsible biometrics  |  standards