This paper investigates the physical layer security of active reconfigurable intelligent surface (ARIS) assisted non-orthogonal multiple access (NOMA) networks in the presence of external and internal eavesdroppers. The study derives closed-form expressions for secrecy outage probability (SOP) and secrecy system throughput, considering both imperfect and perfect successive interference cancellation (ipSIC and pSIC). The secrecy diversity orders of legitimate users are obtained at high signal-to-noise ratios. Numerical results confirm that ARIS-NOMA networks outperform PRIS-NOMA and ARIS/PRIS-OMA networks in terms of SOP and secrecy throughput. The study also reveals that increasing the number of active components and amplification factor does not necessarily improve security due to the presence of thermal noise, and that both thermal noise and residual interference affect secure communication. The research highlights the robustness of ARIS-NOMA networks in handling both external and internal eavesdropping scenarios.This paper investigates the physical layer security of active reconfigurable intelligent surface (ARIS) assisted non-orthogonal multiple access (NOMA) networks in the presence of external and internal eavesdroppers. The study derives closed-form expressions for secrecy outage probability (SOP) and secrecy system throughput, considering both imperfect and perfect successive interference cancellation (ipSIC and pSIC). The secrecy diversity orders of legitimate users are obtained at high signal-to-noise ratios. Numerical results confirm that ARIS-NOMA networks outperform PRIS-NOMA and ARIS/PRIS-OMA networks in terms of SOP and secrecy throughput. The study also reveals that increasing the number of active components and amplification factor does not necessarily improve security due to the presence of thermal noise, and that both thermal noise and residual interference affect secure communication. The research highlights the robustness of ARIS-NOMA networks in handling both external and internal eavesdropping scenarios.