As AI systems become more widely adopted, understanding how related risks emerge and evolve is increasingly important. Media-reported AI incidents and hazards offer valuable insights into the types, frequency, and impact of AI-related harms. This paper analyses real-time data from the OECD.AI Incidents and Hazards Monitor (AIM), organising media reports into 14 thematic clusters. It identifies three broad patterns of proportional media coverage – increasing, intermittent, and declining – and tracks how attention shifts across themes such as synthetic media, child safety, cyberattacks, privacy, and health. While some risks are gaining visibility relative to others, certain themes are receiving comparatively less coverage over time. This analysis offers a structured view of media-reported AI incidents and hazards, supporting more informed discussions on the real-world impacts of AI systems.
