Glossary term
Glossary term
Governance and Compliance
The practical arrangement of committees, roles, workflows, policies, control owners, metrics, and evidence that makes AI governance repeatable across teams rather than dependent on ad hoc judgment. The model defines who can approve high-risk AI, who can block release, who owns testing, who monitors obligations, and how unresolved risks reach senior leadership.
Microsoft's three-line model for AI governance has product teams as first line, the Office of Responsible AI as second line, and Internal Audit as third line.
Capital One's Model Risk Management program, based on SR 11-7, includes a Model Risk Committee chaired by the Chief Risk Officer with documented escalation thresholds.
Accenture's myNav Trusted AI operating model defines four pods covering policy, technology, people, and assurance, used to deliver responsible AI programs to Fortune 500 clients.