Abstract: In the web big data environment, human cognition is disturbed by data, especially multimedia data. The threat of multimedia to human cognitive security lies in the false, harmful and cognitive bias information. Based on the theory of human cognition, aiming at the multi-level impact of social network financial media on human cognition, this paper analyzes and studies the technical threats and protection means faced by multimedia cognitive security in different historical periods, and focuses on the multimedia cognitive security technology based on artificial intelligence, including the forgery and forensics of text, audio, image, video and financial media, It builds a key foundation for multimedia cognitive security. Based on the above research, this paper combs and suggests the future research direction of multimedia cognitive security. The intervention shaping of a new cognitive environment based on multi-agent autonomous cooperation will be the core issue of cognitive security research in the future.