Enhancing Data Security: A Hybrid Approach of AI-Driven Steganography and Encryption

Abstract

In the era of technological development and the Internet, the volume of data transmitted in digital networks is constantly increasing. Ensuring data security has become one of the important challenges in our time. Encryption processes protect data security, but they are often exposed and attract attention. Steganography models are a technique that hides sensitive data but lacks cryptographic protection. The study proposes a hybrid security approach that combines encryption strength and data hiding to be secure against digital attacks. The proposed method takes advantage of one of the artificial intelligence techniques represented by deep learning, which depends on dynamically changing weights during encryption and embedding in the image. This allows us to obtain strong security and high imperceptibility. In the proposed approach, security is enhanced through several layers, the first of which is dynamic changes to generate random numbers and variable encryption as a result of the dynamics of the encryption key and finally hiding the data in a way that cannot be detected. The experimental results showed the merit of the proposed approach through the strength of the results such as the uniformity of the histogram peaks and high entropy = 8 and high imperceptibility represented by BSNR = 91dB. Our research contributes to enhancing data security and countering cyber attacks by exploiting artificial intelligence techniques. Future work has been proposed that opens up horizons for studies using other artificial intelligence techniques such as machine learning and improving real-time data processing in the digital network.