Media assets shared via social media can rapidly spread, even if they do not present a true representation of reality. The assets might have been manipulated with photo editing software, artificially created using deep learning techniques or used out of context. At the same time, editing software and deep learning techniques can be used for creative or educational media production. Clear annotation of media modifications is considered to be a crucial element to allow users to assess trustworthiness of media. However, these annotations should be attached in a secure way to prevent them from being compromised. Various organizations have already developed mechanisms that can detect and annotate modified media assets. However, to achieve a wide adoption of such an annotation approach, interoperability is essential. Therefore, the JPEG Committee has initiated the so-called JPEG Fake Media exploration. The objective of this initiative is to produce a standard that can facilitate a secure and reliable annotation of media asset creation and modifications. The standard shall support usage scenarios that are in good faith as well as those with malicious intent. This paper gives an overview of the history in media manipulation, discusses state-of-the-art in media forensics and in the creative industry as well as challenges related to AI-based manipulated media detection methods. In addressing these challenges, the paper introduces the JPEG Fake Media initiative as a provenance-based sustainable approach to secure and trustworthy media annotation.