This chapter examines AI incident reporting as a critical practice for developing human-centered AI systems. Drawing parallels to safety frameworks in aviation and cybersecurity, it explores how documenting and analyzing AI failures supports the creation of safer AI systems more aligned to human values and needs. The chapter surveys current incident reporting initiatives, from crowdsourced databases to intergovernmental frameworks, highlighting approaches that prioritize human needs and values. The chapter identifies human-centered features of effective incident reporting systems, including accessibility across diverse communities, impacts of mandatory and voluntary reporting frameworks, and open data sharing practices. Key challenges to AI incident reporting are discussed, including defining and documenting AI incidents, developing shared standards, epistemic uncertainties, and scaling detection methods. As AI systems become both increasingly used in day-to-day products as well as integrated into critical sectors, establishing robust incident reporting mechanisms represents a crucial step toward AI that prioritizes human well-being and learns from real-world impacts.