Modern technologies have enabled illicit firearms trafficking to partially merge with cybercrime, while also allowing its off-line aspects to become increasingly complex. The online trade of firearms, their components, 3D blueprints and illicit substances carried out by criminals on both the surface Web and dark Web is increasingly difficult to address as a consequence of the exponential growth in the amount of information disseminated on the Internet. On the other hand, law enforcement agencies are confronted with significant challenges that require the development of sophisticated technological solutions capable of processing large volumes of data, identifying relevant information in a timely manner and creating networks of connections between potential criminal groups. This article presents a real-world practical system, namely the CEASEFIRE one, powered by advanced artificial intelligence technologies that can assist law enforcement personnel in addressing the above described challenges.

The Dark Web, known for its anonymity and illicit activities, presents considerable challenges for Law Enforcement Agencies (LEAs) due to the complexity and volume of data generated within it. Online marketplaces on the Dark Web are notorious for facilitating illegal activities such as drug trafficking, counterfeit goods, and weapons sales while using advanced obfuscation techniques to avoid detection. The unstructured nature of data on these platforms and their constantly evolving operations make manual extraction and analysis exceedingly difficult.This paper addresses the pressing need for structured information extraction from Dark Web marketplaces, with a specific focus on firearm-related listings. Traditional rule-based methods have proven inadequate due to their reliance on HTML tags and pattern recognition, necessitating more adaptive solutions. Thus, the application of Large Language Models (LLMs) and Prompt Engineering to tackle these challenges is explored. By leveraging the capabilities of LLMs, this study aims to transform the extraction process into a more efficient and accurate system. Various generative models and prompt formulations are tested, to determine the most effective approach for extracting detailed information such as product specifications, pricing, and seller details.The proposed pipeline involves feeding crawled marketplace pages into a generative model, which then identifies Product Details Pages (PDPs) and consequently extracts relevant information from them. The use of LLMs marks a significant advancement over traditional methods, enhancing the accuracy and comprehensiveness of data extraction. Additionally, this research highlights the effectiveness of prompt engineering in improving information retrieval.This work underscores the critical need for sophisticated tools to monitor and combat illegal activities on the Dark Web, particularly in the context of firearm trafficking. By refining techniques for automated data extraction and applying cutting-edge LLM and prompt engineering methods, this study aims to support LEAs in their efforts to disrupt and dismantle criminal networks and enhance public safety.

This paper provides an analysis and comparison of the YOLOv5, YOLOv8 and YOLOv10 models for webpage CAPTCHAs detection using the datasets collected from the web and darknet as well as synthetized data of webpages. The study examines the nano (n), small (s), and medium (m) variants of YOLO architectures and use metrics such as Precision, Recall, F1 score, mAP@50 and inference speed to determine the real-life utility. Additionally, the possibility of tuning the trained model to detect new CAPTCHA patterns efficiently was examined as it is a crucial part of real-life applications. The image slicing method was proposed as a way to improve the metrics of detection on oversized input images which can be a common scenario in webpages analysis. Models in version nano achieved the best results in terms of speed, while more complexed architectures scored better in terms of other metrics.