Stepping out in public used to make a person largely anonymous. Unless you met someone you knew, nobody would know your identity. Cheap and widely available face recognition software means that’s no longer true in some parts of the world. Police in China run face algorithms on public security cameras in real time, providing notifications whenever a person of interest walks by.
China provides an extreme example of the possibilities stemming from recent improvements in face recognition technology. Once the preserve of large government agencies, the technology is now embedded in phones, social networks, doorbells, public schools, and small police departments.
That ubiquity means that although the technology appears more powerful than ever, the fallout from errors is greater too. Last week, the ACLU sued the Detroit Police Department on behalf of Robert Williams, who was arrested in 2019 after face recognition software wrongly matched his driver’s license photo to murky surveillance video of an alleged shoplifter. Williams is Black, and tests by the US government have shown that many commercial face recognition tools make more false matches of non-white faces.
In the US, government use of face recognition is much less expansive than in China, but no federal legislation constrains the technology. That means law enforcement can mostly do as it pleases. Researchers from Georgetown University revealed in 2019 that Detroit and Chicago had purchased face recognition systems capable of scanning public cameras in real time. At the time, Chicago claimed it had not used that function; Detroit said it was not then doing so.
Nearly 20 US cities, including Jackson, Mississippi, and Boston, Massachusetts, have passed laws to restrict government use of face recognition. Portland, Oregon, has gone further—barring private businesses from installing the technology. Some federal lawmakers have expressed interest in placing limits on face algorithms, too.
The outcome of any federal legislation will be determined in part by the industry selling the technology. An analysis by WIRED in November found that mentions of face recognition in congressional lobbying filings jumped more than fourfold from 2018 to 2019 and were on track to set a new record in 2020.
More Great WIRED Stories