www.truckandtrack.com Summer 2026 Truck and Track 31 DANGEROUS GOODS The world of dangerous goods is governed by complex and constantly changing regulations. Personnel must be trained to assign, pack, consign uplift transport, deliver and unload dangerous goods and that is unlikely to change anytime soon. However, many aspects within these roles can arguably be processed by AI. AI is all around us and encroaching into almost every aspect of our lives. It has many guises and is taking on roles and tasks that have historically been reliant on human intelligence. However, there is a lot of concern in the dangerous goods industry that AI pulls information from generic sources, some of which may not be correct. This has the potential to increase risk - even AI itself seems worried. Enter “AI mistakes dangerous goods” into your search browser and the AI Overview may respond, without irony, with…. “When artificial intelligence (AI) mistakes or misinterprets dangerous goods, the results can be catastrophic—leading to explosions, ship fires, or severe regulatory fines.” A very quick scan through posts on the most popular social media site for business pulls up plenty of examples where socalled dangerous goods “experts” have probably used AI images to emphasise the importance of compliance and reliability in their post. Unfortunately, these images, particularly those of hazard labels, are often wrong - genuine dangerous goods professionals are frequently swift to point out the errors in the comments. There is, however, a serious point to this. Incorrect images on a screen might not matter too much, but if they are copied and used on packaging it could cause delays or even fines for non-compliance. Moreover, if AI is picking up incorrect images, it will also be picking up incorrect text which could easily lead to products being classified wrongly and the mis-declaration of dangerous goods, thereby heightening the risk already posed by the hazardous nature of a shipment. Relying solely on AI can create serious problems and mis-interpretation of the regulations that apply. There is no substitution for the actual regulations when it comes to dangerous goods – and training is mandatory! To get more of an understanding of the pitfalls I asked Labeline’s Dangerous Goods Safety Adviser (DGSA), Terry Cooksey, for his opinion on the accuracy of information that comes from AI. TC: “The world of dangerous goods can be a complex and confusing one at times, regardless of how much knowledge or experience someone may have.” So, what’s the quickest way to try and find information relating to dangerous goods? TC: “For most practitioners in the industry, the preferred way is to refer to the book versions of the dangerous goods regulations, even though digital versions have been around for many years. Increasingly, though, AI is becoming a rapid source of information.” But is AI always correct, and can AI give us the answers we are looking for? TC: “Well, AI relies on information it can extract from other online resources, and the amount of information related to dangerous goods that is available online is sparse to say the least, and accurate information is even rarer – mainly because the actual regulations are not freely available on the internet. Because of this, AI often struggles to find reference information, especially when looking for anything more than basic information. Answers are often inaccurate at best - and completely incorrect at worst. Information can, however, be corrected if the tool is being used by someone who has a reasonable level of knowledge of the subject. But to someone unfamiliar with dangerous goods, this incorrect information may actually look correct, especially if no other information is available to dispute the AI. This has led to inaccurate classification of substances, incorrect documentation, even labels that bear no resemblance to those specified in ADR or in the dangerous goods regulations for other modes of transport. Overall, AI can be a useful tool, but the information it provides should always be checked with a trained person with experience of dangerous goods, or ideally, a DGSA. This will help ensure accuracy, which allows us all to transport, store and handle dangerous goods safely.” Labeline’s DGOffice software is available to generate compliant documentation, speed up the classification process and alleviate errors on dangerous goods documentation. However, in this fast-moving world, Labeline is not standing still, and we have been working on a ground-breaking tool for DG professionals… Something Unique is Coming! Yes, there will be a solution that will make life a lot easier for every dangerous goods professional. Currently in the final throws of its development and testing, this new application will pull information solely from the regulations so its data will not be confused or compromised by content from elsewhere! Examples of incorrect hazard labels generated by AI Can we trust AI in the world of Dangerous Goods? Richard Shreeve, Labeline’s Compliance Manager, explores the world of AI and its role in the transport of dangerous goods Richard Shreeve
RkJQdWJsaXNoZXIy OTE1MTA=