最終更新日:
A comprehensive, centralized catalog of all AI systems, models, and agents in use across an organization, tracking their business purpose, risk level, and ownership.
An AI Inventory is the foundational layer of any governance program (often required by regulations like the EU AI Act). It differs significantly from a technical "Model Registry" (used by MLOps teams to track code versions). The Inventory tracks the context of use: Who owns this model? Is it high-risk? Does it process PII? Has it passed a conformity assessment? Without a complete inventory, organizations cannot assess their exposure to regulatory or reputational risk.
実際の導入事例：
A global retailer discovers during an audit that its marketing team is using three different "free" generative AI tools for customer emails, none of which are documented. By implementing an AI Inventory, the retailer centralizes these tools into a single dashboard, assigns a "Risk Owner" to each, and ensures they all adhere to the company's data privacy and tone-of-voice policies.
