The structural problem that the avatar flood exposes is that content authenticity tools are optimized for individual artifacts — a single deepfake video, a single synthetic voice clip — not for coordinated volume operations where the fakeness is almost beside the point. Whether a given avatar account passes a deepfake detector is less consequential than whether a network of three hundred of them, all posting the same caption before a midterm, shifts what a platform's recommendation algorithm surfaces. The OECD's formal incident classification of the …