Article 14
Human oversight
High riskApplicable from August 2, 2026
Applies to: Provider · Deployer
High-risk AI systems must be designed so that natural persons can effectively oversee them while in use, with oversight measures built in by the provider and applied by the deployer.
Key points
- Oversight measures must be built into the system or implementable by the deployer
- Overseers need to understand capacities, limits, and automation bias
- Ability to disregard, override, or reverse system output
- Ability to intervene or stop the system safely
- Deployers must assign competent, trained, and resourced people
Related terms
This page is a plain-language summary for orientation only. It is not legal advice — the binding text is Regulation (EU) 2024/1689.