This paper proposes a secure integrated sensing and communications (ISAC) framework for multi-user systems with multiple communication users (CUs) and adversarial targets, where the design problem is formulated to maximize secrecy rate under joint sensing and communication constraints. An efficient solution is presented based on an accelerated fractional programming method using a non-homogeneous complex quadratic transform (QT), which decomposes the problem into tractable subproblems for beamforming and artificial noise (AN) optimization. Unlike conventional artificial noise strategies, the proposed approach also exploits AN to enhance sensing while avoiding interference with legitimate users. Simulation results show significant gains in secrecy rate, communication reliability, and sensing accuracy, confirming the effectiveness and scalability of the proposed framework.