Verification policy
How each step of the policy procedure is met: in the tool, with the export field that records it, or in the organisation's own systems.
Placeholder. The verification policy template and the policy brief are being written by Smart Story Labs. Replace this page with the final documents, or link to their PDFs from here.
Policy template: steps and record fields
| Policy step | Where it is met | Export field |
|---|---|---|
| Intake | In the tool | intake, asset.sha256 |
| Assessment | In the tool | claimedContext, scope, markers |
| Corroboration | In the tool (fact-check search) and by the assessor | factCheckQuery |
| Optional external check | In the tool, with consent for each file | aiGenerationCheck |
| Decision | In the tool; reasoning enforced when markers are flagged | decision |
| Review trigger | The tool marks the record | decision.reviewRequired |
| Review sign-off and time limits | The organisation's own systems | – |
| Further verification | The organisation; a re-assessment is linked by the file fingerprint | asset.sha256 |
| Record | The export, optionally signed; filing and retention in the organisation's systems | the whole record, signature |
| Disclosure, after use, oversight | The organisation; records carry the tool build for audits | tool.build |
Policy brief
The brief maps the three-layer, human-decided approach to the OECD AI Principles, UNESCO's Recommendation on the Ethics of AI, the G7 Hiroshima Process Code of Conduct, NIST AI 100-4, the C2PA standard and the EU AI Act's transparency obligations.
To be published here by Smart Story Labs.