By Alexa Back
Modified bills from State Rep. Orlando Ramos, Rep. Dave Rogers, and Senate Majority Leader Cindy Creem, H 1728 and S 92, attempt to strike a balance between law enforcement needs and civil liberties.
Nijeer Parks spent 11 days in jail after police flagged him as a suspect in a shoplifting incident. Parks only discovered the evidence against him after his release: facial recognition technology had flagged him as a suspect.
Robert Williams spent over 30 hours in an overcrowded detention cell. During the interrogation, investigators presented camera footage of a Detroit Shinola watch store, which the surveillance technology flagged as Williams based on an outdated driver’s license.
Parks and Williams attempted to clear their names, President and CEO Rahsaan Hall of the Urban League of Eastern Massachusetts said. “We cannot allow this technology to run rampant because the people that are going to be victimized the most are the people who are sparsely represented in this world,.” said Hall at the legislative hearing.
A 2018 study by Buolamwini and Gebru showed Black women were misclassified in facial analysis algorithms nearly 35 percent of the time. In contrast, the accuracy rate for white men was nearly perfect.
Kade Crockford and Gavi Wolfe of the ACLU of Massachusetts discussed bills H 1728 and S 92. They collaborated with Erik Leanred-Miller, a computer scientist and AI expert from UMass Amherst, and the Boston Police Department.
Sweeping demands from the Boston Celtics in an op-ed submitted to the Boston Globe prompted these changes. The team urged Governor Charlie Baker to reconsider his decision to lift regulations on police use of facial recognition due to its implicit racial bias.
The amended bills now incorporate the following steps necessary for utilizing the technology, as outlined by Crockford:
Obtain a Warrant: The officer must first obtain a warrant.
Submit the Image: The officer must bring the desired image and the warrant to the state police.
Make a Case for Facial Recognition: A case must demonstrate exigency.
Specialized Unit Involvement: If approved, officials within a specialized unit of trained analysis will conduct the search and provide the results to local police officials.
To ensure fair due process and address public concerns, Crockford assured the bills prohibit mass surveillance, emotion analysis, and wrongful arrests.
“In public spaces [mass surveillance] should be off limits and off the table in a democratic, free and open society” Crockford continued.
A current concern involves emotion analysis, wherein software attempts to discern people’s feelings based on facial expressions.
“There are companies that are trying to sell software to police departments that says, you know, ‘we can tell you whether or not someone's lying’” Crockford noted. Amazon Web Services and Springboard Research Ltd. presently sell facial technology that gauges a customer's level of satisfaction or disinterest in specific products.
Miller pointed to Robert Williams to address racial bias concerns: “This potential match for Robert Williams cannot be used to arrest him. It must be corroborated with other evidence. It is not justification for an arrest.” If a case uses facial recognition, the state must present all evidence to the jury and the defendant.
If passed, Crockford noted these bills could spark a broader conversation about the use of facial recognition in the U.S., potentially influencing legislation in other states. “The intent here is to really narrow the possibility for abuse and misuse and increase the possibility for effective accountability and oversight.”
The legislative body has not scheduled a hearing for the bill yet.
Comments