This work develops an approach to solve the integral equations for dynamic electromagnetic scattering problems based on the physics-informed neural networks (PINNs), which were originally proposed to solve partial differential equations (PDEs). Since different from the applications in most of the existing PINN studies, the function to be evaluated here is complex-valued, network structures to handle the phase information contained in the complex-valued numbers are investigated. An adaptive activation function is employed to improve the performance of the PINN solution. Numerical simulations on two-dimensional (2-D) electromagnetic scattering problems have been conducted to validate the proposed method.