AI camera fines in Australia: why more drivers are challenging them

AI cameras are everywhere now

Over the last couple of years, “AI” road-safety cameras have been switched on across every Australian state. They photograph passing vehicles and use software to flag suspected offences — a phone held to the ear, an unfastened seatbelt, speeding, running a red light — at a scale no team of officers could match. The vast majority of these fines are now generated by machines, with people only entering the loop afterwards, if at all.

That scale is exactly why the error rate matters. When a system processes hundreds of thousands of images, even a small percentage of mistakes means a very large number of wrong fines landing in real people’s letterboxes.

They get it wrong — and here’s how

The technology is not as reliable as the fines make it look. A criminal lawyer told carsales that AI cameras are “notoriously unreliable” at detecting mobile-phone use, because the software keeps mistaking ordinary objects for a phone. Reported real-world examples include:

• A wallet, a glasses case or a battery pack read as a mobile phone.
• A driver fined for “holding his phone” who was actually holding his sunglasses case — he proved it on review by submitting a photo of the case and of his dashboard phone mount, and the roughly $400 fine was dropped.
• In bright sunlight, a seatbelt blending into a driver’s clothing in the black-and-white image, which the software read as no belt.

Seatbelt cameras have caused particular controversy because many fines relate to passengers — including children — wearing a belt incorrectly. Drivers have pointed out they can’t safely lean over and check how a passenger is wearing their belt while driving, yet they’re the ones penalised.

Authorities say a flagged image is reviewed by a qualified officer before a fine is issued. The lawyer who spoke to carsales noted there’s often “no way of confirming” how thorough that check really is — which is part of why so many fines fall over once someone pushes back.

What to check in the camera image

Before you pay, view the photo on the issuing authority’s portal and run through this quick evidence-analysis checklist (it’s also in our free printable guide below):

• Is the “phone” actually a phone? Wallets, glasses and sunglasses cases, battery packs, food and even your own hand have all been mistaken for one. Can you see a screen, or just a dark shape the software guessed at?
• Where’s your phone? If it was in a mount or out of frame, that undercuts a “held phone” reading.
• Is the seatbelt really missing? In bright sun or against dark clothing a worn belt can vanish in the image, or sit lost in shadow behind an arm.
• Driver or passenger? Many seatbelt fines relate to a passenger — and you can’t safely adjust someone else’s belt while driving.
• Is it even your vehicle? Check the make, model and colour — and that the plate is exactly yours. Misread and cloned plates do happen.
• Do the details match? The date, time, location and lane should line up with where you actually were.

Authorities are dropping cases

This isn’t just frustrated-driver talk — the withdrawals are showing up in the official numbers. In Western Australia, the road authority reported pulling 6,114 infringements for 3,413 drivers in the cameras’ first seven months. When drivers actually asked for a review of AI seatbelt fines, around 60% were withdrawn, wiping more than $1 million in penalties, and prompting a formal review by the state’s Road Safety Commission after it emerged the cameras were issuing more than $1 million in fines a week.

The courts are weighing in too. In a Queensland case widely reported in early 2026, a driver successfully challenged an AI-issued seatbelt fine. The magistrate found the prosecution hadn’t proven he knowingly allowed a passenger’s seatbelt breach, and accepted it was reasonable for him to believe his passenger was buckled up properly. The fine was thrown out — and legal commentators said others in a similar position could follow.

The pattern is consistent: when a camera fine is wrong and the driver pushes back with evidence, a meaningful share of those fines don’t survive.

What this means for you

If a camera fine has landed and something about it feels off, the takeaway from all of this is simple: it’s worth a closer look before you pay. Two things are firmly in your favour:

• Asking for a review is free, and it does not increase your fine. If the review is declined, you simply owe the original amount.
• The image is evidence that can be wrong — and you can usually view it on the authority’s portal and judge for yourself.

None of this is a guarantee. The authority makes the final decision, and plenty of camera fines are perfectly valid. But a wrong one is far more common than the official letter would have you believe, and the people who challenge them — calmly, with evidence — are the ones getting them withdrawn.

How to challenge an AI camera fine

The path for most drivers is a free internal review with the issuing authority — not court. Here’s the shape of it:

• Look at the image. View the photo on the authority’s portal. Is the “phone” actually a phone? Is the belt visible? Is it even your vehicle?
• Gather evidence. Photograph the object that was mistaken for a phone, your phone mount, your permit — whatever supports your account. Do it before anything changes.
• Request a review within the window on your notice (commonly 28 days). State plainly why the fine is wrong, point to your evidence, and ask clearly for it to be withdrawn.
• Keep it factual. Don’t make admissions you don’t need to. A focused, evidence-led letter beats an angry one.

That last step is exactly what Fight My Fine does for you. You answer a few plain-English questions, we give you a free read on how strong your grounds are — and tell you honestly if they’re weak — then draft an editable letter addressed to the right authority for your state and fine type. See how to dispute a mobile phone fine, a seatbelt fine, or a speeding fine in your state.