On the one hand, the technology of data deep synthesis, which has been used in thousands of industries, improves the production efficiency, improves the quality of life, and helps the economic and social development. On the other hand, many malicious applications also seriously threaten personal rights, political security, social stability and national security. This paper starts with the concept and classification of data deep synthetic, combined with the characteristics of data to be proved and the current criteria of the accepting, focuses on the key risks in affecting the effectiveness of authentication algorithm, causing trust decline, infringing personal rights and interests, and threatening the security system. Then, it put forward a series of methods, such as strictly curbing source violations, reshaping the expert witness, and strengthening the supporting mechanism of circumstantial experience.