Source Register
| ID | Source | Role in the original mapping | Review status |
|---|---|---|---|
| S-001 | NIST, TEVV-Athlon Framework for Evaluating AI Systems, initial public draft NIST AI 200-2 (Aug. 2026); draft DOI | Initial external assessment-framework case; Events, Tools, Blocks and assessment construction | Overview verified; section-by-section review pending |
| S-002 | NIST AI Risk Management Framework, AI RMF 1.0 (2023) | Risk-management context and TEVV terminology | Overview verified; compare relevant provisions |
| S-003 | NIST, Building Evaluation Probes into Agentic AI, project page (2026) | Evidence-grounding probes, structured audit trail and agentic evaluation instrumentation | Project overview verified; inspect linked code and papers |
| S-004 | NIST AI 800-4, Challenges to the Monitoring of Deployed AI Systems (Mar. 2026) | Post-deployment monitoring categories, barriers and open questions | Publication metadata verified; full report review pending |
| S-005 | PAgE 2026: Principles of Agentic Engineering, PLDI 2026 | Specifications, testing, verification, monitoring and repair for agents | Workshop overview verified; identify exact relevant paper(s) |
| S-006 | Apollo Research: Claude Sonnet 3.7 (often) knows when it is in alignment evaluations (2025) | Evaluation awareness as a possible validity threat to behavioral measurements | Research overview verified; methodology and limitations review pending |
Source-level work needed
- Record stable citation and document version; retain local references to relevant passages with page or section numbers.
- For NIST AI 200-2, verify the exact wording of sections identified in the earlier comment outline against the initial public draft.
- Identify which PAgE papers, if any, are directly relevant.
- Distinguish the NIST probes project's stated scope (grounding and source-support checks) from broader claims about all agentic AI assurance.
- For evaluation-awareness work, distinguish observable expressions of awareness from actual awareness, as the latter may be underdetermined.
Relevant formal research, not an external framework
- Formal Theory: Structural Assurability: definitions, claim resolution and proof obligations.
- Paper 320: Structural Assurability: manuscript development.
The external source register should stay independent of theory-derived interpretations; crosswalks belong in study-specific mapping records.