A Queensland man has successfully overturned a seatbelt fine after challenging the accuracy of AI-powered traffic cameras, setting a potential precedent for thousands of similar cases across Australia.
The ABC reports that the magistrate ruled the AI system's determination that the driver was not wearing a seatbelt was unreliable, noting that the technology's error rate had not been properly disclosed to the court.
The case centered on a $1,033 fine issued after an AI camera system flagged the driver as not wearing a seatbelt. The driver insisted he was wearing one and that the AI had misinterpreted shadows and clothing as evidence of a violation.
Mate, this is what happens when we let algorithms issue fines without proper oversight. A thousand-dollar penalty based on a machine's best guess, with the burden of proof somehow landing on the driver to prove the computer wrong.
The magistrate was particularly critical of the prosecution's failure to provide data on the AI system's false positive rate. Without evidence of the technology's accuracy, the court ruled it could not be satisfied beyond reasonable doubt that the offense occurred.
The ruling has significant implications for Australia's expanding use of AI-powered traffic enforcement. Multiple states have deployed these systems, issuing tens of thousands of fines annually for seatbelt violations, mobile phone use, and other offenses.
Legal experts say the decision could open the door to challenges of existing fines if governments cannot provide robust evidence of their AI systems' accuracy. The Queensland government has reportedly issued over 50,000 AI-generated seatbelt fines in the past 18 months.
Transport and Main Roads Minister Bart Mellish defended the AI camera program, stating that it has been "rigorously tested" and that the vast majority of fines are accurate. However, he acknowledged the department would review its procedures for providing technical evidence in court.
