This paper investigates physical layer security (PLS) in a full-duplex (FD) system assisted by movable antennas (MAs). The system features an FD base station (BS) with multiple MAs for transmission and reception, serving an uplink (UL) and downlink (DL) user, each operating in half-duplex (HD) mode with a single fixed-position antenna (FPA). A single-FPA eavesdropper (Eve) is present, and artificial noise (AN) is used to obstruct Eve's interception. The objective is to maximize the sum secrecy rate (SSR) of the UL and DL users by jointly optimizing the BS's beamformers and the positions of MAs. An alternating optimization (AO) method is proposed to solve the non-convex optimization problem, decomposing it into three subproblems and solving them iteratively. Simulation results show significant performance gains in SSR compared to benchmark schemes. The AO method combines successive convex approximation (SCA), semidefinite relaxation (SDR), and particle swarm optimization (PSO) to achieve a locally optimal solution. The proposed system outperforms existing schemes in terms of SSR, demonstrating the effectiveness of MA-assisted FD communication in enhancing security and performance.This paper investigates physical layer security (PLS) in a full-duplex (FD) system assisted by movable antennas (MAs). The system features an FD base station (BS) with multiple MAs for transmission and reception, serving an uplink (UL) and downlink (DL) user, each operating in half-duplex (HD) mode with a single fixed-position antenna (FPA). A single-FPA eavesdropper (Eve) is present, and artificial noise (AN) is used to obstruct Eve's interception. The objective is to maximize the sum secrecy rate (SSR) of the UL and DL users by jointly optimizing the BS's beamformers and the positions of MAs. An alternating optimization (AO) method is proposed to solve the non-convex optimization problem, decomposing it into three subproblems and solving them iteratively. Simulation results show significant performance gains in SSR compared to benchmark schemes. The AO method combines successive convex approximation (SCA), semidefinite relaxation (SDR), and particle swarm optimization (PSO) to achieve a locally optimal solution. The proposed system outperforms existing schemes in terms of SSR, demonstrating the effectiveness of MA-assisted FD communication in enhancing security and performance.