The increasing levels of criminal media being shared in peer-to-peer (P2P) networks pose a significant challenge to law enforcement agencies. One of the main priorities for P2P investigators is to identify cases where a user is actively engaged in the production of child sexual abuse (CSA) media – they can be indicators of recent or on-going child abuse. Although a number of P2P monitoring tools exist to detect paedophile activity in such networks, they typically rely on hash value databases of known CSA media. As a result, these tools are not able to adequately triage the thousands of results they retrieve, nor can they identify new child abuse media that are being released on to a network. In this paper, we present a new intelligent forensics approach that incorporates the advantages of artificial intelligence and machine learning theory to automatically flag new/previously unseen CSA media to investigators. Additionally, the research was extensively discussed with law enforcement cybercrime specialists from different European countries and Interpol. The approach has been implemented into the iCOP toolkit, a software package that is designed to perform live forensic analysis on a P2P network environment. In addition, the system offers secondary features, such as showing on-line sharers of known CSA files and the ability to see other files shared by the same GUID or other IP addresses used by the same P2P client. Finally, our evaluation on real CSA case data shows high degrees of accuracy, while hands-on trials with law enforcement officers demonstrate the toolkit’s complementarity to extant investigative workflows.