A False Match, an Arrest at Home
Detroit police arrested Robert Williams in front of his wife and young daughters after facial recognition wrongly matched his photo. His lawsuit produced some of the strictest police rules on the technology in the US.
A match is a guess. What has to happen before a guess becomes an arrest?
What happened
In January 2020, Detroit police arrested Robert Williams at his home in Farmington Hills, Michigan, in front of his wife and two young daughters. A facial recognition search had wrongly matched an old driver’s licence photo of him to surveillance footage of a 2018 theft from a Detroit shop. He was held for nearly 30 hours.
Who spoke up
Robert Williams and his family. With the American Civil Liberties Union he sued the city of Detroit in 2021, and that July he gave testimony to a subcommittee of the US House Judiciary Committee about the arrest and what it did to his family. He has continued to speak publicly about it.
What changed
In June 2024 the case settled, with binding rules for Detroit police. They may not arrest anyone based only on a facial recognition result, or on a photo lineup run straight after one. They may not put a facial recognition lead in a lineup without independent, reliable evidence. Officers must be trained on the technology’s risks, including its higher rate of misidentifying people of colour, and the department must audit cases going back to 2017. The court keeps jurisdiction for four years to enforce the agreement. The AP reported that the city would also pay him $300,000.
What to keep straight
- The $300,000 figure comes from AP reporting, not from the settlement text.
- This is a story about policing, not school. It is here because it happened to a family, and because it shows what goes wrong when a machine’s match is treated as proof.
Talk about it
- The software produced a lead. People turned it into an arrest. Where should the checks have been?
- His daughters watched the arrest. Who else is affected when a system gets one person wrong?
- The new rules require “independent, reliable evidence.” What should count?
Go further
AI Education Needs Authority Literacy Before Agent Literacy
Agentic AI is shifting from answering questions to taking actions. Before students learn to make AI agents more capable, they need to learn how to define, inspect, and govern what those agents may access, decide, and do.
Read the article Teach it Classroom Kits · Ages 11–15The Human Sorting Machine
Write a rule before you see who it sorts. Then meet the people it excluded, and one thing about them the rule couldn't see.
View the kitSources
- Written testimony of Robert Williams, House Judiciary Subcommittee on Crime, Terrorism, and Homeland Security, US House of Representatives (13 July 2021).
- Civil rights advocates achieve the nation's strongest police department policy on facial recognition technology, ACLU (28 June 2024).
- Final order of dismissal and settlement agreement, Williams v. City of Detroit, No. 21-10827, US District Court, Eastern District of Michigan (hosted by ACLU) (June 2024).
- Detroit paying $300,000 to man wrongly accused of theft, making changes in use of facial technology, Associated Press, via ClickOnDetroit (28 June 2024).
More accounts to read
All Community Voices-
Families
A Risk Score on the Hotline
A county scores child-welfare calls using public records. Research found racial gaps in its recommendations, and parents with disabilities believe it flagged them.
2016–2023 · Pennsylvania, US Read the account -
Families
The Benefits Scandal That Brought Down a Government
A Dutch risk model that weighed citizenship helped wrongly brand tens of thousands of parents as fraudsters. Parents pushed back, and the government resigned.
2021 · Netherlands Read the account -
Students
The Summer an Algorithm Set the Grades
With exams cancelled, a model lowered 39 percent of teacher-assessed A-level grades, and small classes fared best. Students marched; the government reversed within days.
2020 · England Read the account
Summarised from the public record. Follow the sources for the full story.