An AI-Based Legal Advisory System for Traffic Violation Verification Using Official Traffic Regulations

Automated traffic enforcement solutions are widely
implemented nowadays in order to enhance road safety and
efficiency of enforcement procedures. However, independent ver
ification of traffic violation notices by law is barely investigated.
Current studies also show that manual inspection of traffic
violations is a time-consuming and error-prone process [1],
whereas AI-powered enforcement systems now enable detecting,
classifying, penalizing, and notifying authorities about violations.
However, the reviewed literature lacks sufficient information
about the tools and methods that can be used to check whether
the recorded violation reason and the imposed penalty are
consistent with the existing traffic regulations from the legal
perspective. The importance of addressing this problem becomes
obvious taking into account that among the problems that traffic
rule digitization faces is the question of consistency with the
original purpose of traffic rules [2]. In order to fill this gap, the
proposed research introduces a solution to this problem based on
AI-assisted legal verification. In the case of AI-camera notices, the
framework is meant to authenticate both the reason behind the
violation in question and the fine imposed; in case of traffic police
receipts, where publicly available proof of violation would not be
available, the focus of verification will be on the amount of the
fine. There is also motivation from the field of legal-AI research
which argues that legal advice provided by an LLM comes with
certain dangers due to the necessity of expertise involved and
serious consequences [3]; there is also motivation from the RAG
area which stresses the importance of maintaining fidelity to
provided sources [4]. Furthermore, the framework also includes
a blockchain-based integrity system to maintain authenticity of
verification reports produced by the framework.