The idea is to use AI in traffic cameras to automate the application of sanctions, but the pilot phase of the project ended up generating an unexpected problem: thousands of records have to be reviewed before they can be turned into fines. And there are those who have to prove that they are not an elephant.
Athens is testing a new Digital Road Offenses Certification System, which aims to gradually replace handwritten sanctions with a procedure fully digital.
In theory, the objective is to accelerate process management, reduce administrative burden and reinforce road safety through tools capable of detecting risky behavior on the road.
It’s a good idea, in theory. Difficulties arise when technology passes into practice, in real traffic.
According to the Greek newspaper, during the pilot phase, the percentage of errors or incorrect records will have reached between 90% and 95%. Of the 5500 alerts generated by the system, only 400 were considered correct after being analyzed by the Greek police.
The functioning of the system does not mean that each image captured by a camera automatically becomes a sanction.
First, the cameras identify a possible infringement; This information is then validated by the competent authority. Only if this control passes is that the digital fine can be issued and notified to the driver.
This sequence helps you understand why the problem is not just in the possible disputes presented subsequently by citizens, but also in the volume de material that the administration has to analyze.
According to the data that Ta Nea had access to, 1300 records associated with the alleged use of cell phone and 3800 related to speeding were discarded.
The system has been in testing since the end of March and uses two main sources.
On the one hand, cameras installed on buses of OSY, the public operator of urban road transport in Athens, designed to detect improper circulation or irregular parking in BUS runners.
On the other, a smart camera network linked to the Ministry of Digital Governance, focused on infractions such as driving without a seat beltnot wearing a helmet, running red lights or use your cell phone while driving.
The problem with looking inside the car
The lawyer Tryphon Tsoumanis explained to Ta Nea that drivers may be forced to defend themselves against impossible records.
“There are cases like this and citizens are in the difficult position of trying, in a contestation procedure before the police bodies, prove that they are not elephants…The basic problem with AI-enabled cameras is that they lack the ability to weighting”, he stated.
The transport specialist Stavros Konstantinidis highlighted that cameras can be useful in improving road safety, especially in the case of external infractions. However, he warned that behaviors inside the vehicle raise further doubts.
“Running a red light and exceeding the speed limit are reliably recorded offences. The problem is those committed inside the vehicle.o, such as not wearing a seat belt or talking on a cell phone,” says Konstantinidis.
By May 30, 2026, according to the Greek newspaper, the new mechanism had generated 2453 digital fines. 420 challenges were presented, equivalent to 17.12% of the total, and 52 were accepted, corresponding to 2.11% of all sanctions issued. Many were related to technical failurespoorly readable data, timing discrepancies or exceptions associated with seat belt use.
Shadows, colors, camera angle or even an object such as a cigarette can alter the reading, notes .
The Greek experience shows the challenge of applying AI to control traffic without turn every dubious image into a burden for citizens or for the administration itself.
