Silent cells — coverage gaps across the catalog

Silent count per topic

For each topic, the number of catalogued instruments whose operative text the editorial review records as silent. Rows where more than half the catalog is silent are highlighted amber — these are the topics with the thinnest cross- jurisdictional coverage. Sorted by silent count, descending.

Amber rows: more than 50% of instruments are silent on the topic. Editorial caveats apply — see “What silent does and doesn't mean” below.

Silent count per instrument

For each instrument, the number of catalogued topics its operative text the editorial review records as silent. Instruments with narrow scope (declarations, framework principles, sub-federal acts) naturally cluster near the top; instruments with broad scope (omnibus statutes) cluster near the bottom. Sorted by silent count, descending.

Amber rows: more than 50% of topics are silent for the instrument. A high silent count is not, by itself, a criticism — narrow-scope instruments (sub-federal acts, single-topic statutes) are designed to be silent on most of the catalog. Read these counts alongside the instrument's scope statement on its catalog page.

Notable absences

Hand-curated absences with clear policy salience. Each narrative names a real (instrument × topic) cell in the catalog; none are fabricated. The entries are validated against the live COVERAGE map at render time — if the editor reclassifies a cell out of “silent”, the entry drops out rather than rendering a stale claim.

• G7 Hiroshima AI Process Code of Conduct × AI-Driven Worker Displacement

  The G7 Hiroshima AI Process Code is silent on AI-driven worker displacement. The Code's 11 commitments centre on developer-side risk management (red-teaming, transparency, content provenance) rather than the labour-market consequences of frontier-AI deployment. By contrast, OECD AI Principles §1.1 names inclusive growth + workforce, and Brazil's PL 2338 codifies worker-rights provisions. Frontier-lab voluntary commitments and the AI Safety Institute network share this gap.

• G7 Hiroshima AI Process Code of Conduct × Compute + Model-Weight Export Controls

  The G7 Hiroshima AI Process Code is silent on compute + model-weight export controls. The controls regime operates on a parallel track (US BIS rules, EU dual-use Regulation 2021/821) outside the standard AI-governance instruments. The silence is structural: G7 members coordinate on export controls via separate ministerial fora rather than as part of the Hiroshima Process.

• EU AI Act × Sovereign AI Doctrine

  The EU AI Act does not articulate a sovereign-AI doctrine. The Act's framing is risk-based + market-internal; sovereign-AI debates (state-level model autonomy, national-champion subsidies) operate through industrial-policy channels (the AI Continent Action Plan, InvestAI) rather than the AIA itself. This makes the AIA's silence on sovereign_ai an editorial-policy choice, not an oversight.

• OECD AI Principles (Recommendation) × Compute-Threshold Reporting

  The OECD AI Principles (2019, updated 2024) are silent on compute-threshold reporting. The Principles predate the post-GPT-3 compute-thresholding regime (EU AIA Art. 51 10²⁵ FLOP, US EO 14110 10²⁶ FLOP). Subsequent OECD work (the Hiroshima Process Reporting Framework, AI Compute Trends report) operates outside the Principles' formal recommendation text.

• Bletchley Declaration on AI Safety × Biometric Identification

  The Bletchley Declaration is silent on biometric identification. The Declaration's scope is deliberately narrow — frontier-AI catastrophic + bio-risks — and excludes deployment-context topics (employment, healthcare, criminal justice, biometric ID) that other instruments treat as central. The silence is a frame-choice rather than a coverage gap: a regulator citing Bletchley on biometric ID is misreading the text.

What “silent” does and doesn't mean