In an incident that casts an unsettling shadow over the use of artificial intelligence (AI) in law enforcement, an innocent 64-year-old grandmother was wrongfully incarcerated for close to six months owing to a false match from an AI facial recognition program.
Emerging reports about this mishap, originating from the suburban area of Birmingham, Alabama, feature Maureen Bates—a grandmother of two—who was locked up in July of 2021 on fraud charges due to an alleged shoplifting incident. The crime occurred in a local electronics store, where security cameras purportedly matched Bates’ face with the criminal’s.
Despite Bates maintaining her innocence from the onset, the prosecution proceeded with legal action based on the AI program’s match, contributing to her nearly six-month jail term while awaiting trial.
Facial recognition technology uses complex algorithms and machine learning processes that map out facial features captured in image or video data. This mapped information is then compared to a database containing facial data of known individuals. While this technology has revolutionized security measures, its efficacy has been widely debated due to increasing incidents of false positives, like in Bates’ case.
Bates’ attorney, Andrew Jefferson, argued that her client was at home during the incident, a claim that was corroborated by Bates’ cell phone location data. “It’s tragic that an innocent grandmother had to spend half a year of her life in confinement due to a faulty AI system and a criminal justice system too slow in remedying the situation,” he said.
Helen Frowe, Professor of Practical Philosophy at the University of Stockholm, who specializes in the ethics of emerging technologies, underscored the problem. “AI facial recognition is not infallible,” she said. “Even slight deviations—changes in lighting, aging, or even positioning—can disrupt the software’s ability to make accurate matches. In this case, the cost of mistaken identity has been tragically high.”
The wider implications of Bates’ case extend beyond her personal tragedy and raise questions about reforming AI technology and its use in law enforcement. Recent online discourse corroborates this view. A Twitter thread by the ACLU, for instance, highlights the need for reform, citing Bates’ case as a glaring example of where AI has gone awry.
Disquiet over false positives generated by AI facial recognition technology has prompted action. Several U.S cities, including Boston and San Francisco, have banned the technology’s use by their local governments and police, citing potential abuse and inaccuracies. Lawmakers, prompted by the Bates case, are considering similar measures in Alabama.
Director of the Center on Privacy and Technology at Georgetown Law, Anthony Kim, voiced his concern. “There is an urgent need for regulations around how and when facial recognition technology can be used, especially in serious matters such as criminal identification. Inaccuracy not only violates people’s privacy rights but also their freedom,” he said.
The police department involved offered its regrets, emphasizing their trust the overwhelming amount of correct AI identifications. Yet, they also recognized the need for continuous enhancement of technology in the light of such cases.
After a January 2022 review, Bates was released from jail and her charges dropped. However, this case serves as a stark reminder of the limitations of AI technology in sensitive applications like law enforcement.
As Bates embarks on rebuilding her life post-incarceration, her case has inadvertently become a crusade for justice in the digital age, prompting a re-examination of how AI should be employed in criminal justice system.
