# AI Impact Assessment Template

## System summary
- Name / owner:
- Vendor / model / version:
- Intended use:
- User-facing or internal:
- Consequential decision area:
- Input data and output/action:
- Human review path:

## Assessment
1. Legal basis, notice, consent, opt-out, and appeal requirements.
2. Bias/fairness testing and affected groups.
3. Accuracy, reliability, limitations, and foreseeable misuse.
4. Privacy leakage, prompt injection, adversarial input, and security risks.
5. Training/fine-tuning/embedding retention and deletion.
6. Vendor restrictions on training, secondary use, subprocessors, retention, and incident notice.
7. Health/FDA, finance/FCRA/ECOA, employment, housing, education, children, and Colorado/California ADMT review.
8. Monitoring metrics, incident thresholds, rollback plan, and retirement plan.