Trustworthy AI and the governance of misinformation: policy design and accountability in the fact-checking system

Purpose This study aims to examine the governance of artificial intelligence (AI)-enabled fact-checking systems as sociotechnical infrastructures that shape the production, validation and contestation of truth in digital public spheres. It identifies regulatory approaches and institutional mechanisms that foster transparency, accountability and epistemic legitimacy in the algorithmic governance of misinformation. Design/methodology/approach Adopting a conceptual and policy-oriented approach, this study draws from policy studies, regulatory theory and AI ethics. It analyzes emerging governance models across diverse jurisdictions and international organizations, situating fact-checking systems within broader epistemic and normative contexts. This study also follows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) protocol to structure and transparently report the selection process of reviewed documents and policy sources, ensuring methodological rigor and replicability. Findings AI-based fact-checking systems are not neutral; they embed assumptions about knowledge, authority and evidentiary standards. Effective governance demands inclusive, adaptive frameworks that prioritize procedural justice and public accountability. This study highlights innovative institutional designs such as participatory audits, multi-stakeholder oversight bodies and algorithmic inoculation strategies. Research limitations/implications The findings provide actionable guidance for policymakers, regulators and platform designers seeking to enhance the trustworthiness of AI verification systems. By emphasizing procedural transparency, participatory governance and epistemic inclusion, the framework informs policy development that balances innovation with accountability, supporting the design of AI systems aligned with democratic values and public trust. Originality/value This study develops a multidimensional framework for governing algorithmic truth by integrating legal mandates, sociotechnical infrastructures and epistemic norms. By reframing AI fact-checking as a contested epistemic infrastructure, it contributes to critical debates on algorithmic accountability, epistemic justice and the democratic governance of truth in AI-mediated societies.