The invention relates to an AI face-changing video detection method based on a multi-task learning model, which belongs to the field of computer vision and deep learning, and comprises the following steps: pre-training a model based on multi-task learning to detect a modified face-changing video, and positioning a modified area for each query, the model is an automatic encoder and comprises a Y-shaped automatic decoder. A semi-supervised learning method is used for improving the generality of the network, valuable information is shared before multiple tasks, the sum of an activation loss function, a segmentation loss function and a reconstruction loss function is reduced, and an optimizer is used for optimization, so that the performance is improved. And for video input detection, averaging is performed on the probabilities of all frames to obtain the probability that the input is true or false. According to the method, AI face-changing video detection can be improved.