AI Clearance Supported Tools & Models
Canonical out-of-box AI catalog used by AI Clearance for enterprise governance: coverage taxonomy, risk levels, per-entry metadata, and freshness pipeline outputs.
Table of Contents
1) Methodology and 90% coverage rubric
Coverage is weighted by practical enterprise usage concentration (not list size). AI Clearance prioritizes high-impact categories: chat assistants, coding copilots, enterprise copilots, and API model providers.
Target: 90% weighted coverage. Current estimate: —.
Catalog version: —. Generated: —.
Coverage assumptions and category weights are stored directly in the catalog artifact and audited in automation outputs.
2) Risk rubric
Each entry has a deterministic default risk level and rationale. Risk levels map to governance defaults in AI Clearance policy templates.
3) Coverage snapshot
4) Supported tools/models by category
Entries include tool/provider, status, support mode, risk level, common usage, and governance/auth surface.
Loading catalog…
5) Governance guidance
- Managed: Suitable for standard approval policies with recertification and bounded expiry.
- Tracked-only: Allow only with tighter pilot controls and explicit data restrictions.
- Discovery gap: Default deny; require exception path with security sign-off.
- Use alias matching to reduce ambiguous tool naming during intake and reconciliation.
- Prioritize recertification for high/critical entries and any source-stale entries from refresh reports.
6) Review cadence and automation
- Daily: source heartbeat + stale-entry audit.
- Weekly: source signature baseline update + deterministic diff report.
- Automation outputs review-ready artifacts under predictable paths (`artifacts/`, `state/`, `out/`).
- Escalate if high/critical entries are stale or if key source quality degrades.
7) Source citation bundle
Primary vendor documentation and trusted market usage reports are linked below.
8) Changelog
- 2026-02-18: Introduced machine-readable AI tool/model catalog with risk rubric, source citations, alias coverage, and refresh automation artifacts.