Automating the process of selecting countermeasures through multi-criteria decision analysis with large language models

The escalating spread of fake news threatens societal trust, institutional stability, and public safety. Law Enforcement Agencies must respond swiftly to disinformation but face severe constraints in time, expertise, and resources. This paper presents an artificial intelligence-based decision-support tool that applies multi-criteria decision analysis through the analytic hierarchy process, combined with large language models, to evaluate disinformation cases and rank context-sensitive, operationally feasible countermeasures defined by experts. Evaluations with Law Enforcement Agencies in the European FERMI project produced highly rated and robust recommendations across diverse scenarios, demonstrating strong alignment with strategic and operational needs. By automating a complex, traditionally manual task and implementing it to form a decision model, the proposed approach enables scalable, sustainable decision-making to combat digital disinformation in high-stakes, resource-constrained environments.