San Jose homeless detection AI sparks privacy, inequality fears

Occurred: March 2024

Plans by the city of San Jose, California, to implement a new artificial intelligence system to detect homeless encampments caused concerns amongst privacy and human rights advocates.

The city of San Jose has been piloting using car-mounted cameras and AI software to scan for, amongst other things, trash, graffiti, and vehicles that people are living in, and illegal encampments.

According to city authorities, city staff could respond by sending outreach workers to small encampments before they become large ones if the technology was fully deployed

However, the accuracy of the system has proved (pdf) low, with 10-15 pecent accuracy for lived-in cars and 70-75 percent for RVs. Furthermore, critics expressed concerns about potential privacy violations and misuse of the system to criminalise or displace homeless people.

This initiative underscores the delicate balance between addressing homelessness and safeguarding civil liberties.

Operator: City of San Jose  
Developer: SenSen AI; Zyrex
Country: USA
Sector: Govt - municipal
Purpose: Detect homeless encampments
Technology: Computer vision; Object recognition
Issue: Accuracy/reliability; Human/civil rights; Privacy
Transparency: Governance; Marketing