Agentic Fact-Checking

Abstract

The integration of AI into fact-checking has significantly enhanced the speed and scale of misinformation detection across digital platforms. AI-driven fact-checking systems can now access information in real-time, leveraging machine learning, natural language processing, and large language models. However, the same AI systems that enhance speed and scalability in verification also introduce new ethical and technical concerns. Biases embedded in training data, opaque algorithmic decision-making, and inconsistencies in human oversight can undermine fairness, accuracy, and accountability. Without robust safeguards, AI fact-checking may unintentionally propagate misinformation, exacerbate societal biases, and erode public trust in automated verification. This chapter examines the epistemic challenges AI fact-checking faces, namely bias, fairness, and accountability, against the backdrop of accelerating misinformation. It traces the evolution of verification technologies, evaluates the risks of static governance models, and proposes adaptive, participatory epistemic infrastructures as essential to sustaining public trust. It calls for a shift from technical optimization toward democratic and ethically anchored verification ecosystems.