The live facial recognition technology trial conducted by Foodstuffs North Island has been deemed successful and compliant with the Privacy Act. However, the privacy commissioner has expressed concerns.
The trial was conducted across 25 supermarkets from February to September 2024.
Privacy Commissioner Michael Webster noted that since the facial recognition software was developed overseas, it had not been tested on the New Zealand population, making it uncertain whether technical biases existed, including possible adverse effects on Māori and Pacific peoples.
He also stressed the need for improvements such as continuous technology reviews and updated system settings before the facial recognition technology could be permanently implemented or extended to additional stores.
The commissioner initiated an inquiry to assess the privacy implications of facial recognition technology, verify its compliance with the Privacy Act, and evaluate whether it is more effective at reducing serious retail crime than less privacy-intrusive alternatives.
For Foodstuffs North Island’s Julian Benefield, the trial showed facial recognition technology can help reduce harm while still respecting individuals’ privacy.
The primary privacy safeguards Foodstuffs implemented during the facial recognition trial included the immediate deletion of images that did not result in a positive match, the system being configured to identify only individuals who had engaged in seriously harmful behaviour, prohibition of staff from adding images of children, young people under 18, or vulnerable individuals to the watchlist, and the verification of all match alerts by two trained staff members to ensure accuracy.
Other security measures were:
- No sharing of watchlist information between different stores
- The alert threshold for facial recognition matches was increased from 90% to 92.5% likelihood to reduce false positives.
- Access to the facial recognition system and data was restricted to authorised, trained personnel only.
- Collected images were not used for training data purposes.
- The system was continuously reviewed and improved during the trial to address any misidentifications or errors.
“Our trial prevented more than 100 serious harm incidents, including assaults, with a 16% reduction in harm being demonstrated,” Benefield said.
“We welcome the feedback on areas for improvement and will carefully consider their recommendations, including the need to monitor accuracy, before we make any decisions about future permanent use.”
“The trial findings will help other businesses to ask the right questions about whether facial recognition technology is necessary and appropriate for them and to understand what they would need to do to set facial recognition technology up and run it in a privacy-protective way.”
Meanwhile, Consumer NZ chief executive Jon Duffy expressed satisfaction with the Privacy Commissioner’s inquiry.
“We will be reading the inquiry report with interest in the coming days, particularly the strength of the evidence put forward to convince the regulator that the technology has prevented in-store harm events before they occurred,” Duffy said.