The article presents a theory on how accountability boundaries in AI ecosystems can affect organizational structures and decision-making processes. It highlights the importance of accountability assets in ensuring that AI outputs remain legitimate and assignable, despite the modularization of capabilities.