Abstract: The rapid development and application of deep forgery generation technology have posed potential threats to personal privacy data, social stability, national and enterprise security, etc. Starting from the actual business scenarios of power grid enterprises, and aiming at the insufficient generalization ability of existing deep forgery detection methods, this paper proposes a deep forgery detection method based on multimodal information fusion to provide support for protecting the legitimate rights and interests of power grid enterprises. The method proposed in this paper is verified and evaluated in the benchmark face forgery dataset FaceForensics++(FF++). The experimental results show that the method has good performance and robustness in face forgery detection, and multimodal information fusion improves the generalization of the model.