Cameras in the Classroom: Facial Recognition Technology in Schools

August 25, 2020
Claire Galligan, Hannah Rosenfeld, Molly Kleinman, Shobita Parthasarathy

In this report, we focus on the use of FR in schools because it is not yet widespread and because it will impact particularly vulnerable populations. We analyze FR’s implications using an analogical case comparison method. Through an iterative process, we developed historical case studies of similar technologies, and analyzed their social, economic, and political impacts, and the moral questions that they raised. This method enables us to anticipate the consequences of using FR in schools; our analysis reveals that FR will likely have five types of implications: exacerbating racism, normalizing surveillance and eroding privacy, narrowing the definition of the “acceptable” student, commodifying data, and institutionalizing inaccuracy. Because FR is automated, it will extend these effects to more students than any manual system could. On the basis of this analysis, we strongly recommend that use of FR be banned in schools. However, we have offered some recommendations for its development, deployment, and regulation if schools proceed to use the technology.

Key findings

  1. Facial recognition is racist.
  2. Facial recognition brings state surveillance into the classroom.
  3. Facial recognition punishes nonconformity.
  4. Facial recognition companies profit from children's personal data.
  5. Facial recognition is inaccurate.

While we strongly recommend a ban, below we provide policy recommendations if schools decide it is absolutely necessary to implement the technology. In addition, in appendices to the full report we have also provided stakeholders (e.g., parents/guardians, students, and school administrators) with sample questions to help them evaluate the technology.

National Level Recommendations

  1. Implement a nationwide moratorium on all uses of FR technology in schools. The moratorium should last as long as necessary for the national advisory committee to complete its work and for the recommended regulatory system to be fully and safely implemented on a national level. We anticipate that this process, and hence this moratorium, will last 5 years.
  2. Enact comprehensive data privacy and security laws if they are not already in place.
  3. Convene a national advisory committee to investigate FR and its expected implications, and to recommend a regulatory framework to govern this technology. The national advisory committee should be diverse in terms of both demographic and professional expertise. This committee should include experts in: technical dimensions of FR (e.g., data scientists); privacy, security, and civil liberties laws; social and ethical dimensions of technology; race and gender in education; and child psychology. The committee should also include those involved in kindergarten through high school (K-12) operations, including teachers, school administrators, superintendents, high school students, and parents or guardians of elementary and middle school students. Government officials from relevant agencies (e.g., in the US, the Department of Education and Federal Communications Commission) should be invited to participate in the committee as ex officio members; they could provide important insight into the regulatory options available. Representatives of FR companies should be invited to testify periodically in front of the committee, so that their perspectives can be considered in the regulatory process. Finally, efforts should be made to elicit community perspectives, ideally through deliberative democratic efforts.
  4. Create additional oversight mechanisms for the technical dimensions of FR.

State-Level Recommendations

  1. If a state allows FR in schools, it should create programs and policies that fill in any gaps left by national policy as well as establishing new infrastructure for the oversight and management of district-level FR use.
  2. Convene a state-level expert advisory committee to provide guidance to schools and school districts, if a regulatory framework is not created at the national level. There should be a moratorium on adopting FR in schools until this guidance has been provided.
  3. Establish technology offices, perhaps within state departments of education, to help schools navigate the technical, social, ethical, and racial challenges of using FR and other emerging educational technologies. These offices should also provide resources and oversight to ensure that school and district staff are properly trained to use FR technology in a way that is consistent with state laws.

School and School District Level Recommendations

  1. Schools and school districts are directly responsible for the installation and operation of FR, and for any disciplinary action that follows from identification, so they are responsible for most of the oversight actions.
  2. If any alternative measures are available to meet the intended goals, do not purchase or use FR.
  3. Perform a thorough evaluation of FR, including ethical implications, before purchasing it. This is even more crucial in the absence of national regulations or state-level guidance.  What to Ask To assist administrators, parents, guardians and students evaluate specific FR use in their schools, we offer sample questions in Appendices A and B of the complete report.
  4. Develop a plan for implementing the technology before using it.
  5. Do not purchase FR systems that use student social media accounts to improve the technology.
  6. Do not use FR technology to police student behavior.
  7. Delete student data at the end of each academic year or when the student graduates or leaves the district, whichever comes first.
  8. Employ at least one person dedicated to managing and maintaining the FR technology in each school.
  9. Provide regular, age-appropriate guidance to parents, guardians, and students that includes information about why the school has deployed FR, how it will be used, how data will be managed, and what protections are in place to ensure accuracy and equity.
  10. Establish a pilot period and re-evaluation process before full-scale implementation of the technology.