In this era, a massive amount of digital media data is created and transferred every second, in formats including text, photo, audio, and video. Multimedia has been employed in many areas of human society: music, film, and video games make up the majority of our daily enjoyment; medical visuals assist doctors in making more accurate diagnoses; and fingerprints and face photographs are used to identify persons for a variety of purposes. Machine learning (ML)-based technologies significantly improve our capacity to analyze, process, and use multimedia, however, both multimedia processing and ML technologies require a significant amount of computing and storage. Recent advancements in machine learning have led to several safe and resilient applications, such as digital watermarking, digital image processing, speech recognition, and natural language processing. Machine learning algorithms have overcome numerous obstacles; particularly, trained ML models have made it easier for researchers to deliver cutting-edge results. Digital watermark data is scrambled, and a transform domain-based hybrid watermarking approach can be employed to embed the watermark into the transform coefficients. Machine learning-based digital watermarking solutions have recently received a lot of attention, as machine learning-based embedding approaches for digital watermarking allow the watermark to be injected through learning, allowing the extraction algorithm to quickly retrieve the watermark while maintaining invisibility. The goal of this Special Issue is to focus on advances in machine learning-based digital watermarking, as well as future research possibilities. For this Special Issue, we are seeking high-quality original research and review articles on privacy protection and current advances in machine learning-based digital watermarking.