When the advisor is an algorithm: governing agri-food knowledge in the age of AI and misinformation

CONTEXT Artificial intelligence is becoming an important interface through which farmers, advisors and policymakers access agricultural knowledge, while misinformation increasingly circulates through the same digital channels. OBJECTIVE This Perspective develops the concept of agri-food knowledge governance to explain how AI-mediated advice and misinformation create shared challenges of knowledge validation, contextual relevance and accountability. METHODS The paper conceptually synthesises literature on agricultural knowledge and innovation systems, digital agriculture governance, agri-food misinformation and the epistemology of testimony. RESULTS AND CONCLUSIONS The synthesis identifies six governance functions: provenance and traceability, context fidelity, human epistemic intermediation, plural authority and data rights, resilience and literacy, and accountability and redress. These functions shift attention beyond technical accuracy toward locally valid, contestable and accountable agricultural advice. They become especially important as AI systems move from generating recommendations to executing decisions. SIGNIFICANCE Agri-food knowledge governance extends existing agricultural knowledge-system approaches by explicitly addressing epistemic authority, contextual validity and accountability. It provides a framework for evaluating and governing AI-enabled advisory systems while preserving plural and locally grounded sources of agricultural knowledge.