26 February 2026 - According to a new Policy Brief from Tech 4 Tracing, artificial intelligence computer vision represents an emergent technology for the detection of weapons in X-ray and CT systems in airports, customs and parcel post screening. While the pace of advancement of these systems continues to increase, however, their performance depends on large, diverse training datasets that do not yet appear to exist.

Download the Tech 4 Tracing Policy Brief: t4t-policy-brief-apids-2026.pdf

At the same time, evolving regional firearm trafficking patterns require training datasets for so-called Automated Prohibited Items Detection Systems (APIDS) to cover an increasing set of not only complete firearms but also their parts, essential components, and accessories, a growing proportion of which are now fabricated using new materials such as carbon-infused polymers and aluminum used in privately-manufactured firearms.

According to “From Data to Detection: AI-enhanced Weapons Detection in X-Ray and CT Systems”, the growing complexity of the firearms trafficking threat also has implications for the validation and certification of computer vision detection algorithms in APIDS. In particular, it will require increasingly diverse and larger benchmark testing datasets that keep up with the evolving criminal firearm materials, shapes and designs.

In addition, the Policy Brief suggests that multi-lateral bodies such as the European Union have potentially roles to play in fostering the development of robust AI computer vision training datasets to improve firearm detection algorithms in scanning systems. The authors also draw attention to the need to maintain attention to the human-machine loop and to ensure that incremental improvements to detection algorithms do not lead to over-reliance on technologies that have yet to show sufficiently robust performance.