Facial Recognition Technology: Evasion Tactics, Judicial Scrutiny, and Regulatory Gaps

Facial Recognition Technology: Evasion Tactics, Judicial Scrutiny, and Regulatory Gaps

#GS-2 #GS-3 #Governance & Social Justice #IT & Computers #Artificial Intelligence #Good Governance #Transparency & Accountability #Rapid Fire CA #Quick Facts For Prelims

Key takeaways

  • Modern Facial Recognition Technology (FRT) uses Artificial Intelligence (AI) to convert unique facial features into numerical templates for 1:1 verification and 1:n identification.
  • Scientific research shows FRT algorithms can be bypassed with up to a 98% success rate using specially designed makeup patterns.
  • The Supreme Court of India mandated that police FRT use must satisfy the proportionality test established under the 2017 *K.S. Puttaswamy* Right to Privacy judgment.
  • Despite the passage of the Digital Personal Data Protection (DPDP) Act in 2023, India lacks a specific legal statutory framework governing police FRT watchlists, matching thresholds, and data retention.
  • Inaccurate FRT algorithms pose risks of false positive matches and biased arrests, potentially infringing on citizens' fundamental rights under Article 19.

Why in News

  • The debate over facial recognition technology (FRT) and surveillance evasion gained renewed attention recently.
  • This discussion arose following suggestions that creative makeup techniques could help individuals evade automated facial recognition cameras during public protests.

Understanding Facial Recognition Technology

  • Facial Recognition Technology (FRT) is an advanced biometric system that identifies or authenticates individuals.
  • It performs this task by mapping specific facial features extracted from digital images or video frames.
  • Modern FRT uses Artificial Intelligence (AI) instead of comparing images pixel by pixel.
  • The system extracts unique visual patterns and converts them into a numerical representation called a face embedding or template.

Primary Use Cases of Facial Recognition

  • 1:1 Verification checks an individual's face against a single stored identity record.
  • Examples of 1:1 Verification include unlocking smartphones or verifying passenger identity at airport gates like Digi Yatra.
  • 1:n Identification scans an individual's face against a whole database containing many facial records.
  • Law enforcement agencies use 1:n Identification to scan crowds using CCTV cameras to find matches in criminal databases.

Accuracy Limitations and Evasion Tactics

  • FRT systems calculate similarity scores and rely on matching thresholds rather than providing absolute proof.
  • Because a positive match is not definitive proof of identity, it always requires independent manual verification.
  • The CV Dazzle Project in 2010 showed how high-contrast makeup and asymmetrical hairstyles could disrupt automated pattern recognition.
  • This concept took inspiration from World War I naval Dazzle camouflage, which used geometric patterns to confuse enemy targeting.
  • A 2021 study by Ben-Gurion University proved that AI generated makeup patterns targeting key facial regions could fool FRT software with a 98% success rate.

Judicial Scrutiny and Proportionality

  • The Supreme Court of India stated that it will test the deployment of FRT against the principle of proportionality.
  • This requirement of proportionality stems directly from the landmark *K.S. Puttaswamy* Right to Privacy judgment in 2017.
  • The court directed law enforcement agencies to limit real-time FRT deployment strictly to existing criminal databases.
  • This directive aims to prevent indiscriminate and widespread public crowd monitoring by state agencies.

Key Challenges and Concerns in India

  • India enacted the Digital Personal Data Protection (DPDP) Act in 2023 to regulate personal data handling.
  • However, India still lacks a dedicated law governing how police forces deploy FRT systems.
  • There are no explicit legal frameworks covering facial watchlists, matching thresholds, data retention periods, or independent audits.
  • Algorithms frequently generate false positives, especially regarding gender and race, which can lead to wrongful arrests and bias against marginalized groups.
  • Mass surveillance fears and potential state profiling can discourage citizens from exercising their fundamental right to peaceful assembly under Article 19 of the Constitution.