Governance, Law & Compliance
High-risk AI
AI considered likely to create significant consequences and therefore subject to stronger controls.
Example
An employment-screening system receives extra testing and oversight because it is high risk.
Why it matters
The label directs closer scrutiny toward uses where failures could seriously affect people.
You'll hear it
“What makes this use high risk, and what extra checks follow?”
Can you control it?
Not directly
No direct control. This is a category that laws apply to certain uses of AI, not a feature. What an organization builds and how it deploys a system can affect whether that category applies to it; a reader cannot turn it on or off.
Common questions
- Is every large AI model high risk?
- No. The relevant use and framework matter; model size alone does not settle the classification.
- Does high risk mean prohibited?
- Not necessarily. Some frameworks allow such uses when specified requirements are met.
- Can a system become high risk after its purpose changes?
- It may. A new use can bring different consequences and obligations, so the classification needs review.