Walmart AI anti-shoplifting system accuracy, effectiveness
Occurred: May 2020
Can you improve this page?
Share your insights with us
Walmart's AI and camera-based anti-shoplifting system provided by Ireland-based company Everseen has come under fire in the US over its reportedly poor detection rates and tendency to misinterpret innocent behaviour as potential shoplifting.
A group of Walmart workers calling themselves 'Concerned Home Office Associates' circulated a video documenting flaws in the the company's 'Missed Scan Detection' system, including its regular failure to identify unscanned items, and incorrectly identifying personal items as potentially shoplifted.
The system is so poor that Walmart workers have complained about it continuously since it was first introduced in 2017, and call it 'NeverSeen,' according to WIRED.
Databank
Operator: Walmart
Developer: Everseen
Country: USA
Sector: Retail
Purpose: Reduce scanning errors, theft
Technology: CCTV; Computer vision; Machine learning
Issue: Accuracy/reliability; Effectiveness/value; Surveillance
Transparency: Governance
System
News, commentary, analysis
https://www.wired.com/story/walmart-shoplifting-artificial-intelligence-everseen/
https://www.businessinsider.com/walmart-tracks-theft-with-computer-vision-1000-stores-2019-6
https://www.dailydot.com/debug/self-checkout-walmart-tech-stealing-tiktok/
https://mindmatters.ai/2022/07/could-the-self-checkout-ruin-your-reputation/
Page info
Type: Incident
Published: August 2023