Transport NSW has proposed the use of Artificial Intelligence (AI) to streamline the operation of seatbelt and mobile phone cameras within the state.

Under the state’s current system, images captured by the automated system are already processed by AI, with those that meet the criteria for mobile phone use, or incorrect use or lack of a seatbelt, then passed on for human verification before a fine is issued.

In instances where AI does not identify an issue, photos are deleted, and no further action takes place.

1

Changes to the processing would see the human visual verification stage scrapped, resulting in a fully automated process, from the capturing of images, to the AI processing of the collected images, and any resulting penalty notices.

Seven News reports that seat belt and mobile phone cameras alone have generated more than $400 million in revenue over the last five years, with phone cameras in use since December 2019 and seatbelt detection cameras introduced in July 2024.

Critics of the proposal have called out the risk of misidentification of infringements, suggesting that without some level of human processing, the AI-identified fines could result in a repeat of the Federal Government’s ‘Robodebt’ scheme, which was later deemed unlawful by the Federal Court of Australia.

Data published last year revealed that over 15 per cent of seatbelt infringements flagged by AI detection were rejected in the human review stage.

1

Earlier this year, the WA Government, which uses a similar AI-identified detection system, withdrew over 2000 incorrectly issued fines from over 53,000 infringements issued since the system was introduced.

The latest proposal forms part of the NSW Government’s 2026 Road Safety Action Plan, which includes agenda items including evaluation of new and enhanced camera programs, and trials of new technologies to support automated enforcement for what the state identifies as ‘high-risk’ behaviours.

The proposal is yet to be put before the NSW Government for discussion.