AI Data Governance: The Data Chair

Every AI failure is a data failure that graduated; you will take command of your own data estate, govern what the machines are fed, survive the incidents, and leave with evidence that you are the person a company needs before it needs a lawyer. The world is governing models loudly and data silently: model cards get written while the training corpus that made the model dangerous goes unexamined, and the AI era just made the data person the most consequential and most legally exposed role in the building. You work every module as the Head of Data of your own organization (your real one, or the one you know best): you survey the warehouse, the lakes, the pipelines, the shadow spreadsheets sales refuses to give up, the labeling contractor nobody vetted, and the AI systems already in production that were trained on data no one can fully account for. Every module is anchored to a real, documented, citable data or AI failure, and each real event is used exactly once. Decisions persist across modules through the artifacts you keep, and every artifact is attacked: poisoned corpora, red-team auditors, a hostile board, an examiner. No essays exist here. Every mastery gate is an artifact plus a defense, the thing you made and you answering hard questions about it out loud against an AI examiner who has read everything you did: the lineage map, the quality gate, the records of processing, the exposure report, the Article 10 file, the incident log. The living credential does not claim knowledge, it opens the dossier. Prerequisite: AI Literacy and Empowerment or equivalent working fluency with AI tools, plus everyday comfort with data work; Module 0 bridges the gap. AI Governance is a sibling, not a prerequisite.