This study examines the phenomenon of artificial intelligence (AI)–driven disinformation and its implications for public trust within the context of Indonesia’s digital democracy. The research focuses on a deepfake video depicting a figure resembling the Indonesian Minister of Finance, Sri Mulyani Indrawati, falsely stating that “honorary teachers are a burden to the state.” This study employs a qualitative research design using Norman Fairclough’s Critical Discourse Analysis (CDA) framework, which encompasses three levels of analysis: textual analysis, discursive practice, and social practice. The findings reveal that AI technologies are utilized to manipulate meaning through the strategic use of language and visual representation, further amplified by social media algorithms that prioritize emotionally engaging content. Such dynamics accelerate the circulation of disinformation and shape public perception in the digital public sphere. Consequently, the phenomenon contributes to a crisis of public trust (trust deficit) and undermines the legitimacy of digital democracy. By integrating the theoretical perspectives of Information Disorder and Digital Democracy, this study argues that AI-based disinformation should not be understood solely as a technological issue, but also as an ethical and informational justice problem with broader implications for the quality of democracy in the digital age.