Use and misuse of facial recognition technology!

By M.Y.Siddiqui 

Facial recognition technology offers fast identity verification and improved security. It also creates major risks like privacy loss, mass surveillance, and misidentification due to algorithmic bias. It is both a boon and a bane. Benefits of facial recognition comprise speed and convenience, public safety, efficiency and health, and automation. Harms and risks of facial recognition include loss of privacy, insecurity and bias, data securities vulnerabilities, and government surveillance, racial and gender bias leading to wrongful arrests, and non-consensual data collection. These vulnerabilities threaten personal privacy because facial data cannot be easily changed or encrypted once leaked.

Major risks of misuse comprise governments can track citizens in real-time without consent, creating oppressive police states or targeting minority groups or activists. With algorithmic bias, systems often have higher error rates for women and people of colour, which can trigger false-positive matches and unjust police detentions. There is a fear of data insecurity. Biometric databases are vulnerable to breaches exposing immutable personal identifiers that can lead to permanent identity theft. Besides, there is no consent. Public deployment via street cameras or wearable devices strips individuals of anonymity in everyday public spaces.

Governments are regulating facial recognition technology through a patchwork of outright bans on real-time-public surveillance, risk-based classification systems, and data privacy frameworks. Some major countries enforce strict limitations to protect civil liberties; others maintain broad state exemptions, like India, that allow law enforcement wide latitude in deployment. In India, government relies on the Digital Personal Data Protection (DPDP) Act, but critical national security exemptions allow central law enforcement agencies broad discretion to use biometric systems, leading to ongoing legal challenges regarding proportionality and privacy.

Key safeguards and restrictions in the DPDP Act involve bans and moratoriums, prohibiting real time scanning in crowds, protests or public transit hubs. Requires judicial approval or a formal warrant before running targeted police data base searches. Enforces independent audits to minimize algorithmic misidentification rates, which historically skew higher for marginalized groups. It also restricts how long captured biometric templates can be stored on police or municipal servers. This apart, privacy of people, especially, political adversaries, vulnerable sections like marginalized, women, minorities etc. is not safe in India’s democratic fascism.

Despite harmful effects of facial recognition technology, according to official sources in the Union Ministry of Electronics and IT, facial recognition technology fast identifies verification and improved security. It unlocks phones, logs into accounts, and allows touch less entry in seconds without passwords or cards. It helps police find missing persons, track suspects, and secures borders or high-risk areas. It improves efficiency and health by streamlining airport check-ins and provides a touch-free option that reduces germ spread. Organizes personal photo libraries and cuts down on time theft or buddy punching at workplaces. 

Major regulatory approaches include European Union (EU) and United States of America (USA). Under the EU Artificial Intelligence Act, real time biometric identification in public spaces is generally banned for law enforcement, with narrow exceptions for specific threats like kidnapping or terror attacks. Retrospective facial recognition is designated as high-risk and requires strict human oversight and accuracy audits. In the USA, there is no overarching federal law, but regulation happens at state and local levels. Several cities and states have banned or severely restricted police use of live facial recognition, while others require explicit public notice or warrants before scans occur.

However, there are key safeguards and restrictions that include bans and moratoriums, mandatory authorization, bias and accuracy testing, and data minimization to deal with the menace of facial recognition technology. Bans and moratoriums prohibit real-time scanning in crowds, protests, or public transit hubs.  Mandatory authorization requires judicial approval or a formal warrant before running targeted police database searches. Bias and accuracy testing enforces independent audits to minimize algorithmic misidentification rates, which historically skew higher for marginalized groups. Data minimization restricts how long captured biometric templates can be stored on police or municipal services!

Powered by Froala Editor