The development of artificial intelligence should avoid the security risks caused by the technology black box, and it is necessary to ensure the unity of target functions and technologies through the establishment of reviewable, traceable, reputable regulatory mechanisms.