LANDMARK · 6 MIN · STEWARD
Setting AI risk appetite
After this landmark, you can explain why an explicit, written AI risk appetite changes team behavior differently than an implicit or unstated one.
You can still explore it. We’re showing the shared explanation and a related practical view without hiding the knowledge.
Risk appetite (how much risk an organization is willing to accept for a given category of AI use, and where the line sits between ‘fine to try’ and ‘needs review’), defaults to whatever individual teams decide when leadership hasn’t set it explicitly, and teams under delivery pressure tend to set that bar lower than leadership would have.
An explicit, written risk appetite (by use-case category: customer-facing vs. internal, reversible vs. irreversible, low-stakes vs. high-stakes) gives teams a real answer to ‘can we do this’ instead of forcing them to guess, and gives governance a standard to hold deployments against instead of judging case by case after the fact.
This is the canonical concept. It stays the same across learner lenses so personalization never changes the underlying facts.
What this looks like for you
Not your layer directly, but it’s why some companies’ AI features feel carefully bounded and others feel like whatever a team could ship; the difference is often whether risk appetite was set explicitly.
MAKE A DECISION
Two teams at the same company independently decide whether to deploy AI-generated content directly to customers without human review (one says yes, one says no), and there’s no written risk appetite. What does this inconsistency indicate?
CARRY THISFor your organization, check whether a written risk appetite exists for AI use cases. If not, write a rough draft of what the categories (customer-facing, irreversible, high-stakes) and thresholds should be.
- Artificial Intelligence Risk Management Framework (AI RMF 1.0)National Institute of Standards and Technology (NIST) · 2023-01-26
First-pass citations, limited to primary sources; a reviewer will broaden and verify these before this entry leaves draft.