In negative testing, the large state space and feasibility problems cause the difficulty of generating negative test cases from EPN (Extended Place/transition Net)-based mutants. This paper shows the overview of EPN-based mutants, and the details of our negative test case generation technique. In the technique using ACO (Ant Colony Optimization), ants heuristically search better paths (i.e., negative test cases) to find foods (i.e., intended failures) on a field (i.e., an EPN-based mutant).