ACM TechBrief: Facial Recognition
Joshua A. Kroll · ACM eBooks · 2022
Facial recognition use is increasing despite the technology's fundamental limitations, creating profound privacy and ethical challenges. PROBLEM FACIAL RECOGNITION TechBriefsAssociation for Computing Machinery 30Hours in a Detroit jail spent by a Black American man falsely arrested based solely on an erroneous automated facial "identifi cation" from surveillance video. 1 11 Minimum number of European nations using biometric recognition systems in criminal investigations per a Green Party study for the European Parliament. 2 8 Number poised in late 2021 to begin doing so. 3 0 Number of public spaces in which the European Data Protection Supervisor says the use of facial recognition systems is consistent with "fundamental rights and freedoms." 4 25 Minimum number of US states, counties, cities, and other jurisdictions that have banned, suspended, or heavily restricted use of facial recognition systems.5 82.6 Millions of Americans, of a total 331.5 million, who live in those jurisdictions.6 10-100 Number of times more likely facial recognition systems are to falsely identify Black and Asian faces than white faces in NIST study of 189 systems.7 6 Number of US federal agencies, among 24 surveyed by the Government Accountability Offi ce in 2021, using facial recognition technology to generate leads in criminal investigations.810 Agencies found in the same GAO study to be planning to expand its use in some manner.913 Billions of US dollars world facial recognition technology market is estimated to be worth by 2027. 10 FACIAL RECOGNITION: BY THE NUMBERS Facial recognition undermines privacy in new ways, including requiring individuals to avoid public spaces to opt out. Facial recognition is insuffi ciently trustworthy in many high-stakes applications to replace identifi cation by a human. Errors by facial recognition systems continue to disproportionately and adversely aff ect women, minority, and other marginalized populations. Claims that facial recognition technology can be used for purposes other than identifying people, such as emotion recognition, are not scientifi cally justifi able. POLICY IMPLICATIONSACM's Technology Policy Committees provide cutting-edge, apolitical, non-lobbying scientifi c information about all aspects of computing to policy makers in the United States and Europe.To tap the deep expertise of ACM's 100,000 members worldwide, contact ACM's Global Policy Offi ce at acmpo@acm.