Western Australia has seen an 80 percent [1] reduction in road safety offences since the rollout of AI-assisted traffic cameras.
The technology represents a shift in how the government enforces road laws, moving toward automated surveillance to capture violations that previously went unnoticed. While the state reports a significant drop in offences, the system has faced criticism over accuracy and the burden of proof placed on drivers.
Reece Whitby said, "We've seen an 80 per cent reduction in road safety offences since the rollout of AI‑assisted cameras" [1]. These cameras, primarily deployed in the Perth metropolitan area, were introduced in early 2025 to enforce laws such as seat-belt use [2, 3].
Courtney Bembridge said cutting edge traffic cameras are capturing hundreds of thousands [4] of road offences that may have otherwise slipped through [4]. This increased surveillance is projected to deliver a $440 million [1] fines windfall for Western Australia.
Despite the reported effectiveness, the system has produced erroneous penalties. Thousands [2] of fines have been withdrawn after drivers successfully appealed the AI's findings. Some drivers reported receiving fines for not wearing seat belts when they were actually buckled in.
John Doe, a driver who contested a penalty, said he was shocked to receive a seat-belt fine when he was actually wearing one, and the fine was withdrawn after he appealed [2].
The discrepancy between the high volume of captured offences and the number of withdrawn fines has sparked a debate over the reliability of AI in legal enforcement. The government maintains the system is a success due to the overall reduction in violations, while critics point to the thousands of errors as a sign of systemic instability [1, 2].
“We've seen an 80 per cent reduction in road safety offences since the rollout of AI‑assisted cameras.”
The Western Australian experience highlights the tension between the efficiency of AI surveillance and the requirement for judicial accuracy. While the 80% drop in offences suggests a strong deterrent effect, the thousands of withdrawn fines indicate that the technology may still struggle with 'false positives' in complex visual environments. This creates a legal friction point where the state benefits from a massive revenue windfall while citizens must actively contest automated errors to avoid penalties.
