ai-governancelisted

# AI Governance Use this skill to turn responsible-AI principles and applicable requirements into decisions, controls, and evidence across an AI system’s lifecycle. It produces an engineering governance assessment, not legal advice or a certification. Confirm the current authoritative regulation, jurisdiction, sector obligations, contract, and counsel before making a compliance claim. Do not deploy a high-impact or externally consequential AI system based only on a model score, vendor statement, or checklist. Do not hide limitations, remove meaningful human review, use sensitive data without a documented purpose, or treat a model card as proof of safe operation. ## Establish the system record For each use case record: - provider, model/version, owner, deployer, users, affected people, geography, sector, and lifecycle stage; - intended purpose, prohibited or foreseeable misuse, decision authority, autonomy, inputs/outputs, downstream actions, and third parties; - data categories, provenance, consent/purpose, retention, residency, sensitive attributes, labels, and known gaps; - performance, fairness, privacy, security, reliability, latency, cost, accessibility, and environmental objectives; - applicable law/policy/framework, risk tier rationale, decision rights, escalation, approval, monitoring, rollback, and retirement criteria. Maintain an inventory that is versioned and reviewable. Reassess after a model, prompt, retrieval corpus, tool, data, user population, jurisdict