ABSTRACT
As generative AI (GAI) technologies rapidly evolve, the detection of AI-generated, synthetic media alone is no longer sufficient to address the challenges of mis- and disinformation. Attribution and characterization are essential in determining, for example, what news is fake and what is real. This chapter examines the development of detection, attribution, and characterization methods for AI-generated content, grounded in the Theory of Content Consistency and responsible AI practices. Tracing the evolution of detection tools over the past two decades, this chapter presents the argument that explainability and human accountability must be central to future approaches. The rise of sophisticated image, text, audio, and video generation tools has blurred the lines between real and synthetic, requiring systems that not only identify manipulated content but also explain how and why these determinations are made. This chapter advocates for human-in-the-loop (HITL) systems, transparency, and interdisciplinary collaboration to ensure trustworthy, human-centered AI practices. Ultimately, understanding and explaining AI outputs is essential for preserving digital trust in media journalism as well as upholding a collective definition of truth.
