The recent report of the OECD defining Artificial Intelligence (AI) incidents and related terminology closely followed a Corrigendum announcement by the European Parliament (EP) at the mid-April plenary session in Strasbourg, subsequent to the first reading of the forthcoming AI Act. The AI Act includes provisions for managing "serious incidents" involving AI systems. This article delineates the AI incident notification rules within the AI Act, illustrating a two-stage incident notification procedure. Key issues include the challenge of uniformly assessing the threshold for serious incidents, particularly for high-risk AI systems.