REFERENCE · §10 · LAST REVIEWED 2026-04-27

ACF §10Data Governance

Agent data governance covers the firm’s policies and controls for every flow of data into, through, and out of the agent stack — including training-data assertions from providers, prompt-cache retention, customer-data exposure to tools, and the regulator-facing posture for each.

Existing data governance assumes data flows into known systems with documented retention and access controls. Agent stacks rearrange this: customer data enters a foundation model via a prompt, may be cached, may be observed by a tool vendor, and may shape outputs influenced by training data the firm did not select. The framework section maps each flow to the firm’s GDPR / DIFC DPL / CCPA / sectoral rules and produces a defensible posture for each. Specifically for prompt caching — Anthropic’s and OpenAI’s caching behaviour requires explicit governance.

Regulatory anchors

What this covers

Common gaps

Related sections


Take action

Score your firm's readiness across all twelve dimensions with the Agent Compliance Scorecard →

Reference compiled by Sebastian Heine. Editorial perspective at The SHeine Brief.