When seeing no longer means believing: deepfake detection, journalistic heuristics, and ethical verification in Indonesian election reporting

The rapid advancement of generative artificial intelligence has intensified concerns regarding deepfake-enabled political misinformation during election campaigns across Southeast Asia. This study examines how Indonesian journalists recognized, interpreted, and verified suspected deepfake content during the 2024 Presidential Election. Using a qualitative descriptive design, data were collected through anonymous written responses, followed by semi-structured interviews and focus group discussions, and analysed using thematic analysis. The findings indicate that journalists relied on perceptual expertise and heuristic recognition to identify potentially manipulated content under newsroom deadline pressure. However, these initial judgments required systematic verification through source triangulation, contextual analysis, editorial consultation, and technological support before editorial decisions could be made. The findings show that Perception Theory, the Elaboration Likelihood Model (ELM), and the Heuristic-Systematic Model (HSM) provide complementary perspectives for understanding journalistic verification of AI-generated political content under cognitive and institutional constraints. Practically, the study highlights the importance of strengthening institutional verification capacity through robust newsroom protocols, AI-assisted verification tools, ethical governance, and continuous professional training to support journalistic integrity and democratic communication in an era of increasingly sophisticated synthetic media.