In 2013, the U.S. District Court for the Southern District of New York found that the New York Police Department’s (NYPD) stop-and-frisk practices violated the Fourth Amendment, which requires stops to be based on reasonable suspicion. In 2021, the Court appointed a team of scholars at Stanford University, the University of California, Davis, and the University of Michigan to analyze compliance with the ruling by examining footage from NYPD’s body-worn cameras. The goal was to identify key indicators of constitutional compliance in what NYPD officers say and how they say it during encounters with civilians. Now the team has published a report of their findings, which reveal significant racial disparities in officer interactions.
Using AI technology, the research team found that the prevalence of explicit consent language during consent searches was consistently low. The word “search” appeared in only 46 percent of search encounters, the word “consent” in just 13 percent, and confirmatory questions in 21 percent. Interactions in which all three of these key terms appeared represent just 3.2 percent of all interactions analyzed in the study.
The researchers’ language-based models were able to successfully distinguish Level 1 and Level 2 encounters, in which civilians are free to leave, from Level 3 encounters, in which they are detained. Compared to White civilians, Level 1 and Level 2 encounters involving Black and Hispanic civilians were linguistically more similar to Level 3 stops. Unreported stops of Black and Hispanic civilians displayed substantially more linguistic characteristics of constitutional non-compliance. NYPD officers had an increased use of direct commands, casual language, and profanity when interacting with Black and Hispanic civilians.
“Encounters involving Black and Hispanic civilians that are documented as low-level encounters more often resemble detentions in their language, and even during consent searches — interactions that should be voluntary — Black and Hispanic civilians experience higher rates of command language than White civilians,” the authors write. “These patterns suggest that disparities may arise not only in whether stops occur in the first place, but in how they are experienced and communicated once they do.”
