For a high-risk AI system, a quality management system (QMS) is the documented, repeatable operating framework that connects development, risk management, post-market monitoring, incident handling, and regulator communication. section 7.3.3
A working QMS must:
- Cover the system’s lifecycle with documented, repeatable processes. section 7.3.3
- Assign a specific person responsibility for AI quality and compliance; the AI Act does not require a named AI officer. section 7.3.3
- Support prompt investigation and correction when monitoring identifies problems, including restricting or withdrawing the system if needed. section 7.3.3
Responding to discovered bias
If ClaimAssist V5 monitoring reveals systematic demographic bias, the provider must investigate and correct it; if the fix cannot be verified, the system must be taken offline. section 7.3.3
