Do AI road cameras actually make Aussie drivers safer?AI road cameras are catching drivers across Australia, but top experts warn the technology isn't foolproof. Here's why.Ethan Cardinal05:0019 June 20260comments + 3PhotosAs Australian authorities grapple with the country’s rising road toll, numerous government agencies have deployed a range of high-tech AI road cameras that can detect offences related to mobile phones and seatbelts. And while state road bodies remain adamant that these new enforcement cameras are used to help curb the rising death rate, the technology has attracted scrutiny across the country.RELATED: ‘Unreliable’: AI road camera case thrown out by Aussie courtIn May 2026, an NSW court dismissed an AI-detected mobile phone offence after determining that the prosecution could not prove beyond a reasonable doubt that the driver was using his phone illegally.Similarly, the WA state government waived more than 2000 AI-assisted seatbelt offences, totalling more than $1 million in fines.Most recently, a report by the Victorian Auditor General’s Office (VAGO) could not determine whether the state's AI enforcement cameras program – which began rolling out in July 2023 – had contributed to road safety.The VAGO report, published in June 2026, found the Department of Justice and Community Safety (DJCS), which is responsible for the AI cameras, “does not know if or how the program is reducing road accidents, serious injuries and deaths. It does not have a clear approach for how it will measure this". “The department did not collect baseline data on accidents, serious injuries and deaths linked to distracted driving or not wearing a seatbelt before the program started. Current available data does not show if road accidents are caused by these specific behaviours.”In response, a spokesperson for the DJCS reaffirmed the government agency’s position that these cameras are crucial in road safety enforcement.“Distracted driver and seatbelt cameras play an important role in discouraging dangerous behaviour and holding to account those who put the lives and wellbeing of others at risk,” a DJCS spokesperson told Drive.“We’ll continue to assess and evaluate the program as it moves forward to ensure it is delivering for the Victorian community and holding those who do the wrong thing to account.”Despite recent scrutiny, the latest safety data from the Transport Accident Commission – the state's peak road safety body – revealed that 114 lives were lost on Victorian roads as of 15 June 2026, representing a 19.1 per cent decrease compared to the year prior. With their accuracy and effectiveness called into question, how reliable are these new AI road cameras? Here’s what you need to know.+ 3PhotosA new report from a Victorian government agency has revealed it could not determine whether AI road cameras are actually contributing to road safety in the state. Pictured above: Mobile road camera trailer.How accurate are AI road cameras?According to RMIT University transport engineer and professor Nirajan Shiwakoti, AI road cameras are generally accurate in identifying offences, but their effectiveness from a road safety standpoint is still up in the air.“While AI-assisted cameras can perform well in detecting clearly visible offences, we should not confuse accuracy with effectiveness. AI cameras may accurately detect offences, but a separate question is whether they lead to sustained behaviour change and measurable reductions in crashes,” Professor Shiwakoti told Drive.However, he said these high-tech cameras still have some limitations.“Their performance depends on image quality, lighting conditions, camera positioning, vehicle design, and the clarity of the observed behaviour. AI systems are very good at identifying patterns in images, but determining the context and legal meaning of a situation may still require human judgment,” he said.“Another limitation is that these systems focus on specific observable behaviours. Many important contributors to road trauma, including fatigue, impairment, risk-taking and broader environmental factors, are not easily captured through camera-based enforcement alone.”UNSW AI Institute chief scientist Professor Toby Walsh agreed, but added, “Computer vision now largely works with greater than human accuracy. We should, however, be demanding better transparency from authorities fielding these systems to ensure they have been trained to work well enough”.+ 3PhotosSome experts said AI road cameras are generally more accurate than human assessment, though they still require a layer of review, as the technology isn't perfect.How reliable are AI road cameras? Though there have been several instances of AI cameras incorrectly identifying offences, Professor Shiwakoti said this doesn’t automatically mean the technology isn’t accurate.“It’s important to recognise that no enforcement technology is perfect. Isolated examples of withdrawn fines or incorrect detections do not, on their own, demonstrate that a system is unreliable,” he said.AI road cameras deployed in Australia currently work in tandem with third-party qualified human reviewers, and the added layer of assessment aims to reduce the risk of inaccurate penalties. For example, in Victoria, Drive understands that in the three months leading to September 2025, the number of seatbelt infringements issued decreased by 29 per cent compared to the year prior due to the system. While these high-tech enforcement cameras are generally more accurate than human assessment alone, Professor Walsh from UNSW said it'll take time for people to adjust to them. “The bigger issue here is that these automated systems mean that many more people can be caught breaking the law. It will take time for this to be reflected in people's behaviour. In the short term, it means people are more upset as they receive more fines and points,” he said. The chief scientist at UNSW's AI Institute – who was also fined by these AI cameras after his daughter was caught wearing a seatbelt incorrectly – advised checking the proof if a driver has been detected by one of these enforcement cameras. “The systems aren't perfect, and the public should check the evidence and contest when the AI makes a mistake. We need more transparency on the accuracy and errors,” Professor Walsh said.“We should demand evidence that this is improving road safety, that all these extra fines are making roads safer and not just generating revenue for the government.”But not all experts agree.From a legal perspective, a criminal defence lawyer from NSW-based firm, Astor Legal, Avinash Singh, previously said he’s seen numerous instances in which AI cameras incorrectly detected alleged offences and has called on authorities to amend current laws.“As criminal defence lawyers who see numerous examples of erroneous AI-issued traffic fines, we expect to see the laws around AI-detected fines being amended by acknowledging the lower reliability of AI cameras,” Singh told Drive in May 2026. → Add Drive as a preferred source in Google Search 3 ImagesPreviousNext Ethan CardinalJournalist Ethan traded clothing racks for gear sticks in 2023, when he joined the Drive team after freelancing for various fashion and pop culture magazines. Since then, he has delved deep into all things automotive, from emerging social media trends and road rules to industry and consumer news, hard-hitting exposés, and everything in between. Despite a young career in automotive, Ethan's contributions to the Australian automotive industry culminated in him winning the Newspress Rising Star Journalist award in 2024.Read more about Ethan Cardinal