 Concerns about the dependability of digital media are being raised with growing intensity due to the diffusion of easy-to-use editing tools accessible to a wider and wider public. Concerns has reached an unprecedented level with the development of a new class of artificial intelligence tools capable of producing high-quality fake media without requiring any technical know-how from the users. The possibility of using AI to swap the face of an individual from a video to another to create visually plausible fakes, for instance, is causing great sensation due to the possible malevolent use of such technology. All the more that videos have long been considered as the most trustworthy source of information due to the technical difficulties of editing them. The necessity of developing tools to preserve the transworthiness of digital media is the need that our society can no longer ignore. In the last years, several techniques based on deep learning and artificial intelligence have been developed to detect whether an image or a video has been manipulated or to gather information about its history. Despite the promising results achieved so far, however, the application of AI-based media to multimedia forensics presents a number of shortcomings among which the lack of security is a particular serious one, given the ease with which adversarial examples capable of deceiving AI detectors can be generated. The VIP group, which I've led at the University of Siena, is actively working on the development of secure AI-based multimedia forensics techniques. We do so by leveraging on more than 20 years of experience in security applications of image and video processing, including watermarking, steganography, and steganalysis, processing encrypted signals, and multimedia forensics. In our research, we pay particular attention to the use of concepts borrowed from cryptography, like key-based randomization, to secure artificial intelligence detectors. We are also investigating the possibility to resort to information theory and game theory to model the interplay between the attacker and the forensic analyst in such a way to devise the optimal strategies for both parties and derive the ultimate performance of multimedia forensics in an adversarial setting. We are planning to apply the findings of our research to develop practical and secure video forensic detectors to be applied in real-life applications where the presence of an informed adversary cannot be reeled out. In this framework, we welcome applications of researchers in the postdoc phase of their career who wish to apply for a Marie Curie Individual Fellowship project to spend a two-year research period within the VIP group to carry out leading edge research on secure artificial intelligence-based multimedia forensics. I invite prospective applicants to contact me by email at the address barney at dii.unc.it I look forward to working with you on a successful application and on an exciting research program. Thank you for your attention.