The Cost of Possibility: U.S. Law Enforcement Use of Facial Recognition Technology and Violations of Civil Liberties

Abstract

Artificial intelligence (AI) has gone from being a theoretical branch of computer science to an applied branch of computer science, seemingly overnight. The increasing rate of innovation for AI technologies and the rapid deployment of AI across many industries promises solutions to some of humanity’s most complicated problems, but it also poses risk to democracy’s most sacred element – civil liberties. This paper explores the risks that facial recognition technology, a branch of AI, poses to privacy and due process rights when law enforcement uses it as a tool for investigation. This important work, culminating in two risk scorecards, serves to provide guidance as law enforcement entities embrace technology faster than the law can protect citizens and residents of the United States. The first risk scorecard evaluates common law enforcement facial recognition use cases against civil liberties and provides a risk score for each scenario. The second scorecard evaluates the same use cases across corporate ethical principles for building responsible AI, providing a risk score for each scenario. The ultimate goal of this work is to provide technologists, legal professionals, law enforcement and lawmakers with a perspective and recommendations to shape the deployment, use and governance decisions related to facial recognition technology.