Sessions

Globally, 10–28 percent of crop yields are lost to pests, and the contamination of food and feed with mycotoxins can seriously endanger the health of humans and livestock. In addition, the food chain from harvesting through processing and marketing to consumption can face numerous intentional threats, which increasingly take non-traditional, hybrid forms.

Our research conducted within the framework of the DEFENSEFOOD Horizon Europe project aims to explore who commits intentional acts, with what motivation, using what means, and at what point in the food chain, as well as what recurring patterns these characteristics form. To uncover the data serving as the basis for network analysis, we conducted a systematic literature review, in which the screening of titles and abstracts, the evaluation of full texts, and structured data extraction were performed by a process supported by large language models (LLMs) and run under human supervision. The corpus contains 433 cases from 58 publications, with 93 coded variables per case. The unit of analysis is the threat event: one perpetrator or group of perpetrators, one target, and one agent group, regardless of how many samples, victims, or years belong to it. Threat archetypes were identified on an attribute co-occurrence network. An edge is created between values of two different layers if supported by at least three cases and two independent publications, in the presence of appropriate network metrics. Communities were identified using the Louvain algorithm. In the next step, we analyzed association pathways from motivation to consequence. The network of the corpus directed at intentional harm consists of 99 nodes and 262 edges, and splits into six communities. We consider three archetypes well-founded: insider and extortion-based contamination, state or hybrid action against primary production, and ideologically or terror-motivated contamination. Three further patterns—organized crime and cyberattacks, commercial-market disruption, physical manipulation—rely on few publications and are therefore reported only as candidate patterns.

The most frequent motivation for the acts is political (18%), personal grievance or revenge (15%), and strategic-geopolitical (14%). The target is the general consumer base in 40% of cases, and critical food chain infrastructure in 17%. The most frequently affected chain segments are food services / catering (16%), primary production (14%), and retail (12%). In 77% of cases, the source provides no detection signal, which in itself points to a lack of detection capability. Motivation-centered network analysis of intentional food chain threats shows that motivation primarily determines who the perpetrator is, what the target is, and which segment of the food chain is affected by the act, while it barely influences the choice of agent and mechanism used.