Research on DeepFakes to Ensure Social Information Security

Yang Zeng, Jialin Lin, Rui Zhan, Yufei Man, Yu Yang

Abstract: AI face replacement technology - DeepFakes is often maliciously used to tamper with video faces. Deepfakes forges the access rights of face authentication confidential information or gains the trust of information carriers, which will bring great hidden danger to the information security in the field of social engineering. In order to promote the reasonable application of AI face replacement technology, the DeepFakes source code is analyzed to define the key technologies. And investigate the current status of DeepFakes research. Based on the principle of DeepFakes tampering, this paper summarizes and proposes the method of detecting DeepFakes tampering attacks from two aspects of image feature and combining depth learning. The robustness and accuracy of different detection methods are compared through experiments.