It is crucial to guarantee the remote operation safety of the Maritime Autonomous Surface Ships (MASS). In this paper, a cyber-security framework of the anti-hijack system in the swarm of MASSs was proposed for important information hiding based on blockchain and chaotic-steganography to generate the watermark camouflage image and ensure the security and concealment of important information transmitted in the ship Ad hoc network. The swarm of MASSs cyber-security problem was divided into two parts: source security and transmission security. Three modules are utilized to address the problem. First, the cognitive mechanism for multi-ship interaction based on VRF-PBFT was constructed. Subsequently, MASS transmit the key information of the plain-text which were respectively subjected to the Henon-Logistic chaotic mapping function to produce strong pseudo-random sequence information. Finally, the encrypted sequence information was combined with the original image as the input data for the encoder deep neural networks, and the watermark camouflage image was generated by U-net with the down-sampling and up-sampling structures to extract the image features that are adaptable to the original image. At the ship base, the watermark camouflage image with the original image were merged to achieve the chaotic-steganography which is inserted into the video stream for hidden transmission. At the shore base, the decoder neural networks utilized the ST-transfer CNN architecture to retrieve encrypted hidden information within images. Finally, six self-made MASS remote operation systems of Tianjin University were applied to validate the method. Histogram, PSNR, MSE and SSIM analysis of images before and after encryption demonstrates the method’s statistical robustness, ensuring the secret and secure transmission of critical information perceived by MASSs in ship Ad hoc network.