FAQ

Frequently Asked Questions (FAQ)

Q: Is this platform a replacement for our Legal team? A: No. AI Governance Hub is a facilitator. It gathers the evidence (lineage, metrics, approvals) that your Risk and Legal teams need to make a decision. It replaces the spreadsheet, not the lawyer.

Q: Can we customize the Risk Levels? A: Yes. While we default to the EU AI Act standard (Unacceptable/High/Limited/Minimal), the schema is flexible. We plan to support custom risk taxonomies in v0.6.

Q: Does this store the actual Model Weights? A: No. This is a Metadata Registry. Ideally, your heavy weights live in S3/Artifactory/HuggingFace. We store the pointer (S3 URI) and the checksums, keeping our database lightweight and fast.

Q: We work with Classified Data. Can we use this? A: Yes. The system can be deployed entirely disconnected from the internet (Air-Gapped). The data_classification field supports levels like confidential and restricted to help you tag these assets appropriately.

Q: How do we back up our data? A: The entire state of the system resides in the PostgreSQL database. Standard database backup procedures (Snapshots, Dump/Restore) are all that is required to secure your governance record.

Q: What is the specialized "Policy Engine"? A: It is a logic layer that intercepts actions. For example, if a user tries to move a Model Status to "Approved", the engine checks: "Does this model have a high risk level? If so, does it have a human approval note?" If not, the action is blocked. This "Safety Guardrail" is unique to governance tools.