Use of external AI expertise
Description
Processes for engaging external AI specialists, researchers, or auditors to supplement internal capabilities and provide independent review.
Implementation Guidance
Engagement Criteria
Define when external expertise is required: independent model validation for high-risk systems, ethics review by external advisors, regulatory compliance assessment, specialized technical capabilities not available internally, and adversarial testing by third-party security firms.
Selection and Vetting
Establish criteria for selecting external AI experts: relevant qualifications and track record, independence from system developers, confidentiality agreements, and conflicts of interest checks. Maintain a qualified vendor registry.
Knowledge Transfer
Require that external engagements include knowledge transfer deliverables: documented findings, recommendations with implementation guidance, and training for internal teams. Prevent over-reliance on external expertise for core governance functions.
Evidence Requirements
- External expertise engagement policy
- Vendor qualification assessments
- Engagement contracts and scope documents
- Deliverable reports and knowledge transfer records