Facial recognition helped Bunnings spot repeat offenders in 62 stores

Australia's Administrative Review Tribunal found that facial recognition in 62 Bunnings stores was effective in identifying known offenders. It relied on evidence of a theft cut of around 13%.

  • 2 an organization

    The Tribunal relied on expert evidence for Bunnings that the system cut theft by around 13%, and its alerts ran in the high hundreds.

  • Not rated a person

    The Tribunal found that staff had good reason to feel safer, but no assault or injury prevented is counted.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Reported benefit.
Evidence

foundstudiedreportedclaimed

An independent court document supports the benefit.

Compared with

Bunnings stores without facial recognition over the same period. Measured.

The system’s part
The Tribunal relied on an expert for Bunnings who compared theft in stores with and without the system. His evidence was not challenged, and the Tribunal noted that cause and effect is difficult to test. The decision describes face detection with a Gabor filter and matching of facial vector sets, and it does not say whether any part learned from data.
Limits
Not yet checked.
Costs and harms
Not yet checked.
Later
Not yet checked.

Download data Cite Suggest a correction

Benefit type
Reduced Violence
System
Hitachi facial recognition system (FRT) on store CCTV
Made by
Hitachi Kokusai Electric Co Ltd (designed the system, supplied to Bunnings by a third-party supplier)
Used by
Bunnings Group Limited, 62 stores in New South Wales and Victoria
Where
New South Wales and Victoria, Australia
Stage
Retired
Date
, when the source was published
Kind
Protective response
Broad benefit
Identifying Objects, Actions, and Events, Monitoring Processes, Materials, or Surroundings
Collections
Scrutiny and findings, Cross-border records, Safeguards
Last revised
31 October 2026

Sources

  1. Regulatory document, by authors with no tie to the system.

    Office of the Australian Information Commissioner. Commissioner initiated investigation into Bunnings Group Limited (Privacy) [2024] AICmr 230, determination of Privacy Commissioner Carly Kind, 29 October 2024. Findings on APP 3.3, 'Suitability'. AustLII copy in the Wayback Machine.

    I do not doubt that the FRT system was more accurate, practical and cost effective than some of the alternative tools and strategies available to the respondent to take appropriate action in respect of certain instances of unlawful activity

  2. Court document, by authors with no tie to the system.

    Administrative Review Tribunal of Australia, Guidance and Appeals Panel (Deputy President P Britten-Jones, Senior Members A Murphy and T Simon). Bunnings Group Limited and Privacy Commissioner (Guidance and Appeals Panel) [2026] ARTA 130, 4 February 2026 (archived copy from AustLII).

    We find that collecting sensitive information by the use of FRT in conjunction with the other security controls was effective and suitable to identify known offenders and if necessary to monitor them so as to reduce the likelihood of them engaging in theft or violent or threatening behaviour.

Details

AI system

Workflow

CCTV cameras at store entrances captured the faces of people entering. The system converted each face to a vector set and compared it with those of enrolled offenders. A member of Bunnings' investigations team reviewed each match before staff acted.

Follow-up

Conditions for replication

Faces must be visible, since masks or other people blocking the view defeat the match. Offenders must be enrolled in the database for the system to flag them.

Entry details

Sector

Retail security.

Purpose

Hitachi Kokusai Electric built the facial recognition system to match faces captured on CCTV against a database of enrolled persons and alert staff to matches.

Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.

Cite this record

Peter H. Vartanian. (2026). Facial recognition helped Bunnings spot repeat offenders in 62 stores (AIBD-0856). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0856.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)

BibTeX Data Field definitions